[7/22/80]
Historical
. Narrative
The
1960s
Table of
Contents
Page
IV. IBM's System/360 and the 1960s 269
34. IBM 269
a. 3000 Series and the Spread Committee . . . 2T]
b. The Spread Repo? ^nd S/360 278
(i) Price/Pe: orn . 2 and Techno-
logical Leade .^ip — Generally .... 279
(ii) System/360 Component Technology . . . 282
(iii) Single Family for All Applications . . 290
(iv) System/360 Compatibility 296
(v) Emulation 304
(vi) System/360 Software 311
(vii) System/360 Peripherals .' 314
(viii) Standard Interface/Modularity ..... 332
Appendix: Examples of System/360 Uses .... Al
35. The System/360 Commitment 341
a. Preparation for Announcement ; 357
b. 360 's Success and Impact on IBM . . - v ... 367
36. Initial Competitive Responses to System/360 . . 376
a . RCA 377
b. GE . . . .' 379
c. CDC 381
Page
d. Sperry Rand 382
e. Burroughs " . . 383
f. Honeywell 383
g. SDS 383
37. IBM's Responses (1964-66) ..... 385
a. Reduction of Extra Shift Usage Charges . . 385
• b. Memory Improvements 386
c. Tape Drive Improvements 390
d. Disk Drive Improvements 39 3
e. Introduction of the Model 20 . . . . . . . 395
f . The Model 90 Program 401
g. The 360/44 412
h. The 360/67 . . 417
38. Educational Allowances ..... 437
IBM's Educational Support Programs .... 444
39. IBM's Unbundling 451
a. Introduction . . ■. -. 451
b. The Continued Demand For Bundling '
in the 1960s 458
c. IBM's Unbundling Announcement . 462
40. Sperry Rand/Univac 473
a. Univac's Problems in 1964 473
b. The 1108 477
c. The Product Line Task Force 480
Page
d. The 9000 Series 482
e. Univac's Success in the Late 19 60s .... 486
41. General Electric ............ ."V*7 488
a. The GE 400 Series ' 490
b. The GE 600 Series 493
c. Time Sharing 505
d. False Starts 512
e. The Management of GE's Computer Operation . 516
f.- GE's Position in the Late Sixties 523
h. The Ventures Task Force and the Decision
to Disengage „ 533
i. Did GE Lose Money 543
42. RCA 547
a. RCA 3301 .' . 547
b. The Spectra 70 Series 551
(i) Compatibility with System/360 .... 552
(ii) Commercial and Scientific Ability •*
of Spectra 70 Series 558
"(iii) Problems with Spectra 70 Series . .; . 560 *
(iv) RCA Success with the Spectra 70 Series 575
c. RCA Computer Systems Division 1969-1971 . . 578
(i) Changes in Management Personnel
and goals 57 8
(ii) The RCA Series 583
(iii) Computer Systems Division's
Problems — Early 1970s 595
d. RCA's Decision To Sell Its Computer
Business To Sperry Rand 606
Page
e. After the Sale to Sperry Rand 615
(i) RCA's Activities 615
(ii) Sperry Rand's Success with RCA's
Computer Systems Division 616
f. Conclusion 616
43. Honeywell 619
a. The 200 Series . . . 619
b. Problems and Solutions 626
(i) Other Systems 626
(ii) Peripherals . . 62 8
c. Marketing Practices . 630
d. Product and Service Acquisitions
and Expansion 632
44. Burroughs 638
a. Burroughs in 19 64: Problems and Changes .. 638
(i) Reduction of Expenses 640
(ii) Increased Product Development .... 642
b. Computer Development 1964-1969 643
(i) The 500 Systems Family • . 644
(ii) Smaller Computers 650
(iii) Peripherals 653
c. Burroughs at the End of the 1960s 654
45. National Cash Register 657
46. Control Data Corporation (CDC) 670
a. CDC's 6000 and 3000 Series Offering
(1963-1969) 671
(i) The 6000 Series 671
*(ii) The 3000 Series 677
b. CDC's Expansion into Commercial
Data Processing 678
c. CDC's Expanding Peripheral Business .... 682
d. Data Centers 684
e. CDC's Acquisitions (1963-1969) 686
f. Conclusion 690
47. SDS 691
a. The SDS Entry Strategy 69 3
b. The SDS 910 695
c. The Expansion of the SDS 900 Series .... 696
d. The Sigma Series 703
e. The Merger 711
48. Digital Equipment Corporation 713
a. PDP 1, 4, 5 and 7 71S
b. PDP 6 721
c. PDP 8 722
d. PDP 10 : . 727
e. PDP 15 729
f . PDP 11 729
g. Peripherals and Software 7 31
h. Competition 732
49. AT&T . . . 736
Competition for Bell System EDP Business . 737
Other AT&T Competition 746
Page
50. The Emergence of IBM Plug-Compatible
Manufacturer (PCM) Competition 750
a. From OEM to PCM 753
b. PCM Entrants 762
(i) Telex . . 762
(ii) Ampex 765
(iii) Memorex 767
(iv) ISS 775
(v) CalComp/Centu? ^ata Systems 776
(vi) Sanders Assr j.at... . Inc 777
c. PCM Price Competition and Success . . . . . 780
51. Leasing Companies 797
a. An Overview of Leasing Company Operations . 79 7
(i) The Investment Tax Credit ...... 800
(ii) Marketing Costs 801
(iii) Capital Availability and Cost .... 803
b. The History of Leasing Companies in
the 1960s . 805
(i) Leasing Company Growth 80 7
(ii) The Emergence of New Challenges . . . 814
(iii) Diversification 818
c. Some Individual Companies 819
(i) Greyhound 819
(ii) Boothe Computer Corporation 821
(iii) Itel 824
Page
d. • The Effects of Leasing Companies on IBM . . 826
52. Service Bureaus " 831
a. Entry and Growth 831
b. Time Sharing and the "Computer Utility" . . 835
c. Competition 840
53. Software Companies 851
54. The Role of the Federal Government 866
55. Planning for New Products 878
Introduction 878
Tape Drive Development: The 24 20 and
3420 (Aspen) ....'. 885
c. Disk Drive Developments: The 330 (Merlin),
2319 and 3340 (Winchester) . 893
(i) The 3330 (Merlin) 893
(ii) The 2319 and 3340 (Winchester) .... 902
d. New Processor Planning (NS and System/3) . 907
(i) Monolithic Logic and Memory 907
(ii) System/3 911
(iii) Virtual Memory 913
(iv) NS Prices . . . . 918
e. Conclusion 922
56. Growth of the EDP Industry 923
a. Increase in the Number of Users of Computers 926
b. Expansion by Existing Computer Users . . . 928
c. Explosion of New Applications of Computers. 929
d. Improved Price/Performance and Ease of Use. 937
57. Conclusion 944
L ^ V. IBM'S SYSTEM/360 AND THE 1960S
2 ! 34. IBM. The 1960s for IBM was an era of great change,
3 : of great risk and difficulty and most of all a decade marked by the
4 | phenomenal success of IBM's 'System/360. * The 360 story begins in
u
IS
19
20
6; * Several witnesses actively participated in the planning, develop-
ment and execution of the System/360 program. Their testimony
7 , provides us with a useful means of understanding this critical
period in IBM's history.
8
Erich Bloch was the engineering ^manager of Project STRETCH from
g. \ October 19 58 to April 1961, and "was responsible for the circuit
design and systems organization and implementation". (E. Bloch, Tr.
[q \ 91468.) In April 1961, Bloch headed IBM's Advanced Technology Study
Committee, which was established to recommend the appropriate logic
H I component technology for future products. (E. Bloch, Tr. 9149 2.)
From June 1961 to September 1964, Bloch led IBM's development of
[2 i Solid Logic Technology and "was responsible for the development,
design and pilot manufacturing of the SLT family of components and
[3 || packaging and their manufacturing". (E. Bloch, Tr. 9146>*)9.)
l^ ; Dr. Frederick P. Brooks, Jr., hired by I3M in 1956 as an engineer,
I helped to design the architecture of the STRETCH computer. (Brooks,
15 | Tr. 22650-51.) In 1960, Brooks became Systems Planning Manager of
j the Data Systems Division (DSD) and was responsible for developing
tg ■ "the plans and architecture" for the 8000 series. (3rooks, Tr.
:| 22656-57, 22665.) Brooks served as Manager of IBM's New Product
Line project from 1961 until 1964 and was responsible for "think [ing]
through the technological and architectural approach to a total corpo-
rate-wide product line". (Brooks, Tr. 22656-57, 22666-67.) From
early 1964 to the summer of 1965, Brooks was Manager of Operating
System/360 (Brooks, Tr. 22673-74) and headed the design and develop-
ment activities for System/3 60 's programming support. (Case, Tr.
77966-67.)
7T ■] Richard Case, in 1962, was a member of the Advanced Systems i
; : Group which was responsible for the design and development of System/ \
22 ij 360, and personally headed the engineering group which was at that time!
j designing what was announced as the IBM System/360 Model 60. (Case, j
-« ij Tr. 7 2010, 73235-3 8.) During this time frame, Case also served on ■
'j IBM's Architecture Committee (Case, Tr. 72008-09; DX 3538), which !
2, ;i was responsible for developing System/36 0' s architecture. (Case, Tr . j
\\ 72008-09.) Case's function on the Committee was to represent all of !
2* j the 360 engineering groups. (Case, Tr. 72012, 73238.) In 1964 to
~ ; | 1965, Case was Assistant Manager of OS/360 (Case, Tr. 77966-67;
i DX 3538) and assisted Dr. Frederick Brooks (Manager of OS/360) in the I
i
'I -269- \
L i^ V. IBM'S SYSTEM/360 AND THE 1960S
2 ! 34. IBM. The 1960s for IBM was an era of great change,
3 : of great risk and difficulty and most of all a decade marked by the
4 | phenomenal success of IBM's 'System/360. * The 360 story begins in
u
IS
19
20
6; * Several witnesses actively participated in the planning, develop-
ment and execution of the System/360 program. Their testimony
7 , provides us with a useful means of understanding this critical
period in IBM's history.
8
Erich Bloch was the engineering ^manager of Project STRETCH from
g. \ October 19 58 to April 1961, and "was responsible for the circuit
design and systems organization and implementation". (E. Bloch, Tr.
[q \ 91468.) In April 1961, Bloch headed IBM's Advanced Technology Study
Committee, which was established to recommend the appropriate logic
H I component technology for future products. (E. Bloch, Tr. 9149 2.)
From June 1961 to September 1964, Bloch led IBM's development of
[2 i Solid Logic Technology and "was responsible for the development,
design and pilot manufacturing of the SLT family of components and
[3 || packaging and their manufacturing". (E. Bloch, Tr. 9146>*)9.)
l^ ; Dr. Frederick P. Brooks, Jr., hired by I3M in 1956 as an engineer,
I helped to design the architecture of the STRETCH computer. (Brooks,
15 | Tr. 22650-51.) In 1960, Brooks became Systems Planning Manager of
j the Data Systems Division (DSD) and was responsible for developing
tg ■ "the plans and architecture" for the 8000 series. (3rooks, Tr.
:| 22656-57, 22665.) Brooks served as Manager of IBM's New Product
Line project from 1961 until 1964 and was responsible for "think [ing]
through the technological and architectural approach to a total corpo-
rate-wide product line". (Brooks, Tr. 22656-57, 22666-67.) From
early 1964 to the summer of 1965, Brooks was Manager of Operating
System/360 (Brooks, Tr. 22673-74) and headed the design and develop-
ment activities for System/3 60 's programming support. (Case, Tr.
77966-67.)
7T ■] Richard Case, in 1962, was a member of the Advanced Systems i
; : Group which was responsible for the design and development of System/ \
22 ij 360, and personally headed the engineering group which was at that time!
j designing what was announced as the IBM System/360 Model 60. (Case, j
-« ij Tr. 7 2010, 73235-3 8.) During this time frame, Case also served on ■
'j IBM's Architecture Committee (Case, Tr. 72008-09; DX 3538), which !
2, ;i was responsible for developing System/36 0' s architecture. (Case, Tr . j
\\ 72008-09.) Case's function on the Committee was to represent all of !
2* j the 360 engineering groups. (Case, Tr. 72012, 73238.) In 1964 to
~ ; | 1965, Case was Assistant Manager of OS/360 (Case, Tr. 77966-67;
i DX 3538) and assisted Dr. Frederick Brooks (Manager of OS/360) in the I
i
'I -269- \
I j; the 1960-61 time period. As we have seen, by that time IBM
o iiwas marketing more than 15 different processors and at least
3 Ij seven separate lines of second generation computer systems.
(See above, pp. 126-49.) The architecture of those systems was
"quite dissimilar", as was their programming. (DX 4740, Evans,
p. 3925.) Whatever software compatibility there was existed
y ijonly over a very narrow range of processor performance.
il
si!
20
9 |! design and development of System/360 ' s programming support. (Case,
|| Tr. 77966-67, see also Tr. 77977, 77982.) Case was a co-inventor on
10 j| the original System/360 patent. (Case, Tr. 71993-99; DX 3538.)
11 i Bob 0. Evans was Director of Systems Development and Planning
i for DSD from early 1961 to mid-1962 (Evans, Tr. 101269; DX 8031 (Tr.
12 |j 101035)), and initially responsible for "personally evaluating ...
|j the 8000 series" and deciding whether IBM should pursue the project.
13 I (Evans, Tr. 101269.) During this period, Evans also served as Vice-
j| Chairman of the SPREAD Committee. * (Evans, Tr. 101270; DX 140 4A, p. 3
14 i| (App. A to JX 38).) In the 1962 to 1965 time frame, Evans was Vice-
|j President of Development for DSD (DX 8081 (Tr. 101035)), and assumed
15 || "worldwide responsibility for coordination of the development" of
!| System/360. Evans, Tr. 101061.)
16 ji
I! Paul W. Knaplund was Manager of Systems Marketing for the Data
17 :j Processing Division (DPD) in 1960 and was responsible for understand-
ing and informing IBM's product divisions of "the functions and
18 ;| prices necessary for IBM products to be economically attractive to
l! users" . (Knaplund, Tr. 90467, see also DX 9033 (Tr. 90458).) In the
19 ij latter part of I960, Knaplund became Manager of Systems Development
for the General Products Division (GPD) and was responsible for pro-
jecting profitability for and meeting profitability objectives of
.various IBM products, including the 1401 and 16 20 processors and IBM
21 |j unit record equipment. (Knaplund, Tr. 90464-68; see also DX 9033
!j (Tr. 90458).) In 1963 Knaplund was named Assistant Group Executive,
22 j Product and Profit Planning for the Data Processing Product Group DPD
ij (Knaplund, Tr. 90474; DX 9033 (Tr. 90458)), and "was directly involved:
23 ; | in the preparations and discussions that resulted in the System 360
I announcement of April 7, 1964". (Knaplund, Tr . 90474-75.) In the
24 ; j 196 4 to 196 6 time frame, Knaplund assumed further executive respcnsi- ■
■Ibilities as Vice President and Group Executive of the Data Processing .
25 ij Product Group, and as Vice President and Group Executive of the
ji Systems Development and Manufacturing Group, which required him "to
iideal with functional, pricing and schedule issues" relating to System
!! 36 and other oroducts. (Knaolund, Tr. 90468, see also DX 9033
1 (Tr. 90453; . )
; i -270-
In addition, the input/output equipment had been developed
"almost uniquely" for each processor in order to optimize the perfor-
3 jimance of each of the different system types. (DX 4740: Evans, Tr.
4
5
6
7
8
9
10
11
12
13
(Telex) 3925.) The result, of course, was a very limited flexibility
in attachment possibilities. As Evans testified, because peripheral
equipment differed for different families or attached in different
[ways to different processors, customers "had great difficulty in mov-
i
ing even from one member of a processor in one family to another, let
alone moving from one family type to another". (DX 4740: Evans, Tr.
(Telex) 3925-26.) In this regard IBM's computer systems were no dif-
ferent from the computer systems of its competitors - (See pp. 15 6-70,
203-11, 229-37, above.) The result of this situation was that cus-
tomers generally acquired set systems and had very little flexibility
14 jito change their configurations as business demands changed.
\\
15 l! a. 3000 Series and the SPREAD Committee. In the early
il
16 111960s, IBM's General Products Division (GPD) was resDonsible for the
jj
17 Ijdevelopment and manufacture of IBM's small and intermediate systems,
ii
18 I'such as the 1401 and 1620, as well as IBM's disk drives. (Knaplud,
ij
19 l|Tr. 90464-65; DX 13890, pp. 16, 18; see also DX 1404A, p. 10 (App. A
20 | to JX 38).) IBM's Data Systems Division was developing and manufacture
(I
21 lling IBM's larger systems, the 7000 Series, as well as IBM's tape drives
22 ij (DX 4740: Evans, p. 3919; DX 13890, p. 16; see also DX 1404A, p. 10
il
23 n(App. A to JX 38).) DSD and GPD were achieving great success in the
24 ^marketplace with their current lin^s — particularly with the 7090 and
25 11401. (DX 1404A, pp. 81-82, 35, 86, 39 (App. A to JX 38).) In fact,
-271-
1 !
i
i
2 !
i
3 !
4
5
6
7
3
9
10
11
12
13
14
15
16
the 1401, which had been announced in October 1959, was the most suc-
cessful computer system that IBM had ever introduced, with domestic
shipments of more than 1600 by year end 1961. (DX 140 4A, p. 75
(App. A to JX 38); DX 2609B, p. 94.)*
Nevertheless, neither of the Divisions was resting on its
laurels; they were planning for the future. If IBM was to continue
to compete successfully, it would have to commit itself to the devel-
opment of even better products . Such a commitment would require large
financial investments by IBM. T. J. Watson, Jr., IBM's Chairman,
fully understood this requirement and reported the following to IBM's
Managers in an April 24, 1961, Management Briefing:
"[0]ur competition is getting stiff er all the time ....
The best way to meet this competition is to keep our prices
competitive. Prices involve costs and earnings .... We need
constantly to spend large sums in research and development of
new products which will not produce revenue for some years to
come. Without funds for this vital expense, competition would
eventually surpass IBM." (DX 8886, p. 43.)
Thus, within both divisions, improvements and extensions to
17 lithe then current product lines were being developed. At GPD, Engineer 1
IS iling Manager Ernest S. Hughes, Jr. (DX 1399, p. 2 (Tr. 33869)), had
19 j
20 I
21 !
set up two groups of engineers — one to pursue improvements to the
1400 family and another to outline and define a replacement for the
1400 family. (Hughes, Tr. 33915.) At DSD, development was even
22 !i further alone. A machine called the 810 6 had been under design for
ii
23 iisome years and was already under construction within DSD when
24 ii
ji
25 I * w e are aware that DX 26 09B is not in evidence but we rely on it
I! because it represents a sworn response by an IBM executive which
j! reflects information taken from IBM's accounting books and records,
ii
i -272-
1 i
L
M
si
•t
i
Dr. Frederick P. Brooks, Jr., came to the Division in 1960 as Systems
Planning Manager. Thereafter, IBM began to develop the 8106* into a
series of machines called the 3000 Series. (Brooks, Tr. 22771.) By
1961, IBM: had spent many millions of dollars en the 8000 Series
development. (Evans, Tr. 1010 47.)
Despite the relatively advanced state of the 8000 project
and the money IBM had already invested in it, there was "vigorous
debate" within the company over whether the 8000 was the right way to
proceed. (Brooks, Tr. 22665-66.) With the first elements of the 8000
nearing announcement, B. 0. Evans, who at than time was Director of
Systems Development and Planning for DSD, was charged with evaluating
the 8000 to determine whether it was a "leadership" program. (Evans,
Tr. 101045-46, 101269.) Evans was charged by DSD's Group Executive,
T. V. Learson, to get the 8000 into production if it was the right
thing to do or, if Evans thought the 8000 Series was the wrong
approach, to do what was right. (Evans, Tr. 101046.) Evans concluded
that the 8000 Series was "wrong" for a variety of reasons:
One , the family was based on "contemporary transistor tech-
nology" and would not be "far-reaching enough". In Evans' view,
it would have been a "terrible mistake" to build a new family of
machines that could be rendered obsolete by competitive products-
incorporating much better transistor technology that would soon
be available. (Evans, Tr. 101043; see also DX 4773, p. 3.)
' A
* The 3106 was an outgrowth of the STRETCH program. (3rooks, Tr
15 ;22771.)
-273-
1
z
3
5"
S
Two, the 8000 had a "lackluster" plan with respect to
peripherals. (Evans, Tr. 101048-49.)
Three , the 8000 Series was planned to be "a range of five
different machines: a small scientific machine, a small business
machine, a medium to high performance business machine, a higher
speed scientific machine . . . [a] superspeed scientific machine."
j lj (Evans, Tr. 101051.) Evans thought that offering this "collec-
3
9-
10
11
12
13
14-
Is
16 .
tion of differing machines with kind of loose ties ... in
their structure" was "a basic mistake from the user's standpoint".
(Evans, Tr. 101049.)
Although Evans believed that the 8000 Series would be an improvement
over IBM's existing product line and might give IBM a "momentary
advantage" over competition, he recommended its cancellation. (Evans,
Tr. 101049; see also DX 4773, p. 1.) On June 27, 1961, W. 3. McWhirtei
wrote Lear son that I3M's Regional Managers had been apprised of the
reason why the 8000 was withdrawn:
l| "[T]he 8000 Series offered insufficient advances to insure -
^ •] our competitive position at this time — [it] is being replanned
with new technology to provide a major breakthrough . . . ."
(DX 14059.)
13 ;i
*- ji In late 1961, T. V. Lear son, rhen IBM Vice President and
22 i| Group Executive, appointed a task force called the SPREAD Committee
;i
21 j to develop a new plan for IBM's data processing products during the
22 ;| 1960s. (JX 38, p. 2; see DX 1404A, p. 7 (App. A to JX 38).) Its
-- II
" \ Chairman was J. W. Haanstra, Vice President of Development for GPD
24 j and its Vice Chairman was Evans, who had become Director of Systems
^c 1
-~ ; Development and Planning for DSD. (DX 1404A, p. 3 (App. A to JX 38);
i
'j
:i -274-
z
I
2
7 .
DX 3081 (Tr. 101035).) Other members of the SPREAD Committee included
Dr. Brooks and J. W. Fairclough, Manager of Product Development at
IBM's Hursley Laboratory in England,* who had been in charge of yet
Ulf another processor development, the SCAMP.** (Tr. 71984-85; DX 4779.)
The Committee issued a report of its recommendations in December
1961. (JX 38, p. 2.) That report is Exhibit A to the System/360
Stipulation of Fact (JX 38) and is also Defendant's Exhibit 1404A-
The SPREAD Committee recommended "termination of the prolif-
eration of IBM products and the development of a family of compatible
processors which would employ a common technology (Solid Logic
Technology or SLT) , a compatible set of peripherals and a compatible
program operating system". (JX 38, p. 2.) The report and recom-
- || mendations of the SPREAD Committee were accepted by IBM management
o ■«
C !
* The remaining members of the SPREAD Committee included D. T.
Spaulding, Group Director of Product Line for the Data Processing
Group; J. D. Aron, Programming and Technology Coordination Manager
for FSD; W. P. Heising, Programming Systems Planning Manager, Develop-
I ment for DSD; H. Hellerman, staff member, IBM Research; W. H. Johnson,-
i Director of Product Evaluation, Corporate Headquarters; M. J. Kelly,
j Senior Engineer and Technical Advisor for GPD; D. V. Newton, Manager,
i Mathematics and Programming for DSD; B. G. Oldfield, Manager, Systems
I Development for FSD; S. A. Rosen, Data Processing Manager for the
j Queens IBM New York Branch Office; and J. Svigals , Manager, Systems
! Marketing for DPD. (Tr, 71984-85; DX 1404A, p. 3 (App. A to JX 38).)
i ** SCAMP was an experimental computer built in 1960 by IBM's Hursley
- j Laboratory in England. (JX 38, p. 5.) SCAMP ' s control function was
|j implemented by the technique of microprogramming. ( Id. ) Although
2"! the SCAMP project was cancelled in favor of System/360 (DX 4779, pp.
i| 2-3) , Fairclough was able to convince the SPREAD Committee of the
3 j benefits of microprogramming, which the Committee adopted as the
j principal means of implementing control functions in System/360. In
A "\ addition, because microprogramming techniques were better known in
;| IBM's British lab than in the United States labs at that time, design
'5 j of the Model 40 — which was the lead System/360 model in development —
| was assigned to Hursley. (Brooks, Tr. 22806-07.) (The importance of
1 microprogramming is discussed below at pp. 3 02-03.)
I
1 -275-
L ! and the development of the New Product Line (NPL) , which ultimately
2| became System/360,* began in 1962. (JX 38, p. 3.)
3' ! The principal alternative course of action, which the
i
4-1 SPREAD Committee considered and rejected, was the addition of improved
5 !■ successors to the then existing product lines, rather than develop-
i
I
Si ment of an entirely new line. (Case, Tr» 73571.) The one course of
i
7 i action that I3M could not afford to take was simply to maintain the
j
3 j status quo and continue marketing its current products . That much
g. I was plain from the "product survival charts" incorporated in the
IQ ! SPREAD Report. (DX 1404A, pp. 73-91 (App. A to JX 38).) Those
i
t ,
11 j charts "showed that all of the existing products in the IBM product
£2 j line were estimated to have very short lives, that they would be very
13 i
quickly coming out of users' installations . . . [b]ecause other
14 ; systems manufacturers were developing new and better products and
that the evaluation was that all of the existing product line was
very rapidly heading toward being non-competitive". (Evans, Tr.
101271-72.)
According to Paul Knaplund, the "'Product Survival Charts'
15
IS
17
IS
is!
•i
2g :! * The processors included in the April 7, 1964, System/360 announce-
" I ment and their NPL designations are set forth below:
21 1
;i
22 :i
■t
:J
24 i
i
System/360 NPL Designation
2030 . 101
2040 250
2050 315
2060 and 2062 400
i
!
| 2070 501 (JX 38, <f 4, p. 3.)
I -276-
i
2 ;
3J
4-i
fi
i
i
T\
a
II
12 S
13
. .. . projected displacement of both announced and planned to be
announced IBM central processing units (CPU's) by newer products as
users ' needs grew and changed and as new technologies and equipment
features enabled electronic data processing (EDP) suppliers to offer
improved products . Those charts demonstrated . . . that IBM had to
move rapidly ahead with the development, of a new line of CPU's or
else competition would soon displace IBM's EDP business". (Knaplund,
Tr. 90473.)
The survival chart for the 1401 (DX 1404A, p. 75 (App. A to
JX 38)) made the point graphically. This most successful of IBM's
systems, announced only two years earlier, was projected to reach a
peak of installations by 1965 , with installations declining rapidly
thereafter. Projections for the rest of the product line were similar.
t ^; l The charts indicated that if IBM did not introduce new, improved
7 = ! products its entire installed base would be replaced by its competitors
,g j| In the face of these projections, the SPREAD Committee stated the
1T :j need for new products to be developed and delivered by 1965. Accord- -
j- : | ingly, they recommended announcement of the first processors in the
'i
, '! line during the first quarter of 1964. (DX 1404A, p. 57 (App. A to
„ J) JX 38) .)
« T j The SPREAD Report, and the Systems Architecture Group which
;! was responsible for implementing its recommendations, created a
„ :i product plan that went far beyond the recognized competitive need for
1
_ .] new and improved products and set forth a revolutionary concept of a
5C . ; future product family. This concept represented a commitment to a
i
| vision of the future development of the industry far more daring and
i
t
i -277-
/ i
3
g
ia
12
13
L4-
LI far-reaching than any of IBM's competitors ever attempted. The
Z l concept, subsequently embodied in IBM's System/360, held the potential,
3. j if successfully implemented, for enormous business success for I3M
4-'r and also for revolutionizing the EDP industry. It sought not just
competitive success with existing users but a vast expansion of the
number and. types of EDP users and uses . At the same time , the magni-
tude of the commitment — the devotion of virtually the entire business
to that concept — carried with it a risk of staggering proportions .
Both internally and externally, the IBM System/360 program came to be
referred to as a "you- bet your company" undertaking. (Evans, Tr.
101126; see also Friedman, Tr. 50 373; Case, Tr. 73561-62.)
b. The SPREAD Report and S/360. The concept for the New
Product Line (NPL) , which became 360, embodied a number of objectives
including :
(i) the clear assertion of price/performance
and technological leadership;
:j (ii) the merger of business and scientific capabilities
17 il
I in a single family of systems (in fact, the attainment of a
iS ;l
ij series of computer systems that would be an industry leader
13 ; i
! in the performance of all applications , hence the origin of the
i name "System/360" to denote the full 360 degrees of the circle
21 \
] (Evans, Tr. 101129));
(iii) upward and downward compatibility across a broad
23 "I
| family of processors;
21 1
! (iv) a comprehensive set of systems software;
M* St
j (v) compatibility of a wide range of peripherals across
1
1
! -273-
1 i
3-j
4-1
»
si
7 I
3
9-
a
1
2
3
.4
the entire family of CPUs ; and
(vi) the substantial user flexibility attainable from the
resulting modularity of the boxes constituting a 360 computer
system. (DX 1404A (App. A to JX 38).)
Each of the objectives held the promise of greater customer
acceptance and a substantial broadening of the demand for and use of
computers , and each raised its own particular challenges and risks .
The attainment of each objective posed obstacles in development,
design and manufacturing, each of which carried with it the possi-
bility of failure. These objectives, and the manner in which IBM
successfully attained them, are discussed in the following pages.
(i) Price/Performance and Technological Leadership —
Generally. The 8000 Series was cancelled because it would not have
been a "leadership" product for a significant period of time,
either technologically or in a price/performance sense. System/360,
its replacement, was clearly intended to be both. In December 196 2,
•[ T. V. Learson, IBM Vice-President and Group Executive of the Corporate
I Staff, wrote to Evans that IBM's aim was to make the new line "eco-
!} nomical as hell, simple to operate and the best on the market".
.6
.7
3
9 |i
] (DX 4795.) IBM's Chairman wrote in June 1963 that it was important
:j for IBM to "make these machines good enough so they will not be just
II ij
ij equal to competition" , because IBM expected that once they were
2. ';[
;| announced IBM's competitors would " immediate lv try to better them"
3 i
! and "I [Watson] want cur new line to last loucr enough so we do not go
14. II
j in the red". (DX 48 06.)
-279-
L i
2
3
5
6 I
7
a ;
9-
ia
ii
12 i
13
14
15
15 ;i
'I
is -i
19 1
■f
2C; !
21 :j
:i
22 ;
\
23 I
24 1
The price/performance of System/360 turned out to be a
spectacular improvement over IBM's earlier product line. (Rooney,
Tr. 12123-24; Welke, Tr. 17079-80, 17304-05; see also Northrop, Tr.
82711; PX 3638 ») In a memorandum written to Evans and Kennard just
prior to the System/360 announcement/ the Manager of Market Analysis
for the Data Systems Division stated that " [i]t is difficult to
estimate the competitive jolt NPL will create. Never before has a
single announcement obsoleted so much existing equipment at one time"
since "NPL will have an advantage over all existing systems offered
by major competitors". (PX 1099A, p. 1.) In particular, the Model 30,
intended largely to replace the 1401, had "six times greater internal
speed" than that system. (DX 3726 (Tr. 78990); see also DX 4740:
Evans, Tr. (Telex) 4034-35; DX4755.)
The following comparisons at the time of announcement illus-
trate these improvements:
Rental Price (with maximum memory)
Maximum Main Memory Capacity
(Chars. )
Performance
(instructions/second)
Performance/Rental Price
1401
$2,680
(DX 573,
p. 6)
4,000
(DX 573,
p. 3)
Model 30
$3,875
(JX 38,
p. 33)
65,536
(JX 38,
p. 32)
5,000
(DX 4740 :
Evans, Tr.
30,000
(DX 4740 :
Evans , Tr . |
. (Telex) 4034-35; (Telex) 4034-i
DX 4755) 35; DX 4755) j
1.87
7.74
-280-
2
3
4
^
a
7 !
3
9-
10 I
11
12
12 !
14
15
16
17 !
13 ]
•I
19 :|
za }
21 :|
lj
22 "»
23?
i
l
.i
24 : i
:!
i
25
Rental Price (with maximum memory ;
Maximum main memory capacity-
Rental per million characters
Performance (multiplications/second )
7090
$43,500
(DX 572A,
p. 6)
Model 75
$60,300
(JX 38,
p. 394)
196,608 (6 bit 1,048,576 (8 bit
characters) characters)
(DX 572A,
p. 5)
$100,708
(DX 572A,
p. 6)
38,200
(Case, Tr.
74220)
.88
(JX 38,
p. 394)
$35,286
(JX 38,
p. 394)
366,000
(Case, Tr.
74220)
6.07
Performance/Rental Price
Knaplund testified that just prior to announcement IBM con-
sidered 360 price/performance to be superior to the best known
competitive systems and substantially superior to the best of IBM's
older computer systems. (Knaplund, Tr. 90503; see also PX 1099A.)
T. V. Learson wrote in July 1964 that System/360 's price/performance
had achieved a 30% to 50% improvement over IBM's previous produce
line. (DX 1525.) Moreover, the analytical methods used at the time
to predict price/performance understated the comparative advantages
of System/360 by failing to take into account the benefits to the user |
stemming from the use of disks, the advantages of compatibility, the !
i
System's improved reliability, the advantages expected to come from its!
software and the availability of large memories. (Knaplund, Tr. 90504- |
05; see also PX 1099-A, p. 1; PX 6204, p. 1.) ;
i
The price/performance and other advantages of System/3 6 j
-281-
i ;
i
z\
a !
i
4-j
si
i
I
s i
7!
3 i
si
10 !
I
I
12. |
13 !
were recognized outside of IBM as well. For example, plaintiff's wit-
ness Frederic G. Withington of Arthur D. Little reported in October
1964 that " [w] ith the introduction of their System 360 equipment, IBM
established the new price-performance standard, for equipment within
the computer industry for the next several years" (PX 4829, p. 16) r
an. opinion which he reiterated during his testimony. (Withington,
Tr. 56591-92.) Withington also testified that System/360, at the
time of its announcement in 1964, was "regarded as the best in terms
of incorporating the most recent developments in systems program and
machine architecture".. (Tr. 56590.) Similarly, in a June 1964
presentation to G. E.'s Executive Office, John Weil called System/360
an "excellent product line with outstanding peripheral offerings" and
stated that it was "no longer possible to offer equipment with a
14 I* significant advantage over IBM". (PX 320, pp. 13-14.) Additionally,
15 I
1
16.1
IT J
is ;
19 ;
RCA's June 1964 "Five Year Plan" noted that System/360 "has and will
have a significant impact on the marketplace and other suppliers are
obliged to meet its capabilities". (PX 243, pp. 5-6.)
(ii) System/360 Component Technology. In explaining his
recommendation to cancel the 8 000 Series, Evans had written: "New
m ~ '1 technology is essential to a new IBM machine family. Committing a
:!
„ T ;j new family's lot to current technology is opening IBM to a major
•! competitive coup". (DX 4773, p. 3.) The improvements in price/
„ ij performance offered bv 360 could not have been achieved without the
!
,| superior circuit technology that Evans had envisioned. (Case, Tr.
-282-
I- i 73244; Evans, Tr. 101048.) Development of such technology (called
2. | Solid Logic Technology or SLT) had already begun in IBM when the
31 ;. SPREAD Committee met. The Committee recommended the use of SLT as
4-1 processor componentry because it "promised improved cost/performance
SJ- and reliability." (JX 38, p. 5; DX 1404A, p. 7 (App. A to JX 38).)
S ; Case testified that "the entire System/360 line . . . was predicated.
7; on the availability of the new SLT technology." (Case, Tr. 72303.)*
8. ; SLT development, which had begun prior to 1961, was acceler-
g ; ated in April of that year on the recommendation of IBM's Advanced
i
1XJ ! Technology Study Committee that a "high priority SLT program" be
JX ■ established^ (JX 38, p. 5.) According to Erich Bloch, who headed
12, \ the Advanced Technology Study Committee until September 1964, that
J3 ; Committee had been charged with recommending the logic component
i
L4 ; technology that IBM should use in its future EDP equipment and with
£5 ■; establishing the schedule and cost objectives for its implementation.
1$ jj (E. Bloch, Tr. 91468-69, 91492, 91686; see also DX 9117, p. 2.)
The Committee decided that the new technology had to be
producible at half the cost of the then current SMS (Standard Modular
^g ■! System) technology and be four times as fast. (E. Bloch, Tr. 91492-
2g it 93.) These performance goals were influenced by both the technology
ij
2«r ;| performance and computer performance that could be achieved by IBM
2^ .] competitors , including both computer manufacturers and component
!i
• I
23 •!
17
18
-a '! * "In fact, in its early states, the Svstem/3 6 oroject was known
Vjas the SLT family." (Case, Tr. 72303.)
?«
-283-
1
2
3
5
5
7
3
3
ia
ii
12
L5
manufacturers. According to Bloch, the existence of such competitors
and their introduction of new products and processes since the 196 0s
has forced IBM to be alert to their offerings in order to remain
competitive in terms of cost, performance, reliability and function.
(Tr. 91690-9-2.)* As T. V. Learson later put it, prior to the intro-
duction of System/360, "IBM had been in the market for a long time
with the old circuitry .... It was time for a change. Competitive
action told us so . . . . " (PX 1900, p. 7.) Because of such
"competitive pressures", the Committee decided that the development
of the new technology had to be accomplished within 18 months and the
delivery of machines incorporating the technology to customers begun
within three years. (E. Bloch, Tr. 91686-87; see also DX 9117, p.
3.)
The Committee considered three courses of action: improve-
ment of the existing technology; development of monolithic technology;
*q ;j and further development of a hybrid technology (SLT) with discrete
T |[ semiconductor components combined with screened circuit elements.
Q ! (E. 31och, Tr. 91492.) SMS was the packaging for discrete components
IS ;|
Tq |i used by IBM in its second generation equipment. (Case, Tr. 72265.)
^ \ It had been designed and developed by IBM for Project STRETCH and was
i
] superior to the discrete component packaging available from outside j
;j suppliers because it was optimized for use in EDP equipment. (E . Bloch;
T
\ * Bloch included such firms as Texas Instruments, Fairchild,
2 ~ :j Motorola, Intel, Mostek, AMD, Hitachi, Fujitsu, Philips and National
.[Semiconductor. (S. Bloch, Tr. 91691-92.)
25 ; j
i
-284-
L
Z
3
4-
Tr. 91486-87.) Despite its contribution to IBM's successful line of
second-generation computers, the Committee concluded that SMS tech-
nology had apparently been pushed close to its limits in terms of
cost, performance and reliability and would not yield the desired
5" If performance improvements* (E. Bloch, Tr. 91493; see DX 9117, pp. 4-
6 i 5, 8.)
7 j, In order to gain additional information about the feasi-
3 ; bility of going directly to monolithic circuitry,* IBM was advised by
g. j other companies, including Fairchild, Texas Instruments and Motorola,
i
IQ ! concerning their development activity with monolithic technologies.
i
l^\ (E. Bloch, Tr. 91694-95.) The Committee concluded that while mono-
£2 i lithics could meet the performance requirements laid down they could
13 || not be produced in the time or at the cost desired. (E. Bloch, Tr.
i
^4 j 1 91492-94; see also Gibson, Tr. 22625-27; DX 4782.) The Committee
75 ; therefore recommended moving part, but not all of the way to mono-
tg j lithics: the continued development of the "hybrid (SLT) configura-
17 i
13 •
tion". (E. Bloch, Tr. 91492-94.)
That recommendation was based on several advantages to the
1Q '! SLT technology: first, it would lend itself well to automation and to
„g ii a fast production buildup; second, it would lend itself "to a product
^ :} spectrum of applications" in processors of all sizes as well as
_,:' input/output devices; third, it would be capable of providing the
j
23 -j
j A ;| * " [T]he total integration of all devices . . . and interconnecting
"^ ;j wiring in a single piece of semiconductor material". (E. Bloch,
- c : ;Tr. 91492.)
1
i
■1
it -285-
necessary speeds or performance ranges; and finally, the semiconductor
packaging would accommodate the semiconductor well, provide the needed
electrical characteristics and give the desired packaging densities -
(E- Bloch, Tr- 91699-701.)
These anticipated advantages were in fact realized and SLT
3 it became a high-performance technology for its day, offering a sub-
stantial increase in speed at a substantial reduction in size.
(Case, Tr. 72301-03; E. Bloch, Tr. 91705; see also PX 3587 (Tr.
25334).) SLT was a "significant advance" in IBM computer technology:
it required less space, power and cooling per circuit than SMS; it had
higher performance and "ten times the reliability" of the earlier
technology — all at a reduced cost. (E. Bloch, Tr. 91496-97; see also
12 ; | McCarter, Tr. 88380; Evans, Tr. 101132.) Thus,. SLT enabled IBM to
?iL ' offer "very substantial gains" in price/performance. (Evans, Tr.
r - ; 101132.) Further, SLT "lent itself to automation" (E. Bloch, Tr.
j
,- !•• 91705) and IBM took advantage of that fact by investing heavily in the
development of automatic tools. (Case, Tr. 72298-301.) "IBM
coordinated the development of tools, the development of a design
I automation system and the production and testing of components with
- !
i
i
5 \
if
7
3
a
ia :
ii
17
IS
19 :»
'] the development of the components themselves. Each of the parts of |
20" ! !
:l the technology took into account the other parts." (E. Bloch, Tr. j
21 ; | j
i 91497-98.) |
22 1 . i
ij Such automation enabled IBM to reduce 'production costs and ■
22 i j
j improve the reliability of its circuits. (Case, Tr. 72301; E. Bloch, ;
2^ ! I
] Tr. 91497.) IBM's "substantial investment" in automatic manufacturing
2= i ■
j techniques was a very important factor in allowing IBM to make System/ i
'i !
i ;
i ;
1 -286- I
360 much more powerful for the same price or to be a lot less costly
for the same, power. (Case, Tr. 72301.) From 1965 to 1969 5LT tech-
nology and the automation which accompanied it gave IBM a cost advan-
tage over other component manufacturers who moved their assembly
outside the United States in order to get a cheaper labor source for
the relatively labor-intensive production processes. (E. Bloch, Tr.
( ;i 91708.) No other computer manufacturer had the equivalent of SLT
technology at the time of System/360' s announcement and. delivery
(Evans, Tr. 101131) , despite the substantial benefits that it held
and despite the fact that SLT was an extension of the existing tran-
sistor technology which was readily available to everyone.
Only with the benefit of hindsight, however, was it obvious
that the SLT decision was the correct one. During the middle 1960s,
a ; up to about the beginning of 1966, criticism of the decision was
- i expressed within IBM. Critics thought that SLT had been the wrong
choice, that by being more aggressive IBM could have gone to mono-
lithic circuits and taken a larger jump forward.* (E. Bloch, Tr.
91695-96.) Implicit in that criticism was the apprehension that IBM
would be the victim of a competitive coup by other companies moving
n \ beyond I3M in circuit development.
"! This failed to happen. Based upon a comparison of the cost
j and capabilities of IBM's SLT circuits with competitive monolithic
\
j i
r
> i
*
j circuits that became available from the mid-19 60s forward, Bloch
13 t
| * 31och also testified that the criticism "died down" when it became
£ I clear that "SLT met all the goals" set for it in a way that could not
I have been done with monolithics. (Tr. 91696.)
-287-
L 1
i
a!
>
4-i
i
5
7!
3
9
ia
11
12 1
12 I
15 ;
i
16 ■;
17
IS
concluded that SLT had as good a performance as those later developed
products, was "much denser" and was produced at lower cost than the
products which IBM's competitors acquired from outside vendors.
(E- Bloch, Tr. 91704-05; see also Withington, Tr. 56591.) Moreover,
when IBM did convert to monolithic circuits in 1968-1970, it was able
to use a great deal of what had been done in SLT to ease the transi-
tion into monolithics. (E. Bloch, Tr. 91698; Dunlop, Tr. 93991.)
This planning for the future had been taken into account by the
Advanced Technology Committee and for that reason IBM designed tech-
niques and tools during the SLT development that could be adapted to
the manufacture of monolithic circuitry. (See E. Bloch, Tr. 91500,
91703, 91494 and Case, Tr. 72300-14 for details of the carryover of
SLT development into monolithics.) SLT still is being used by IBM in
secondary circuit functions of newer products. (E. Bloch, Tr. 91499.)
The advantages of automation, of taking an intermediate step
toward monolithics, and of coordinating circuitry, component and
|[ product development could be fully realized only through in-house
development and manufacture. Accordingly, the Advanced Technology
!i Study Committee recommended the establishment of a components division
20 '"
19 ;|
which would be able to manufacture SLT on a large scale. (E. Bloch,
.„ ] Tr. 91562.)
21 |j .
„ ;! Case called IBM's decision to develop and build its own new
22. *
i • l
ij circuitry "perhaps the riskiest single decision that had to be made by i
■! I
, IBM in the development of System/360". (Tr. 73514.) It required a |
24 I I
?c
substantial capital investment in a new business — developing and manu-
facturing transistor components — in which IBM had had little prior
-288-
4|
ff |
i
s !
7 1
3 |
n I
12 i
i
13 !
experience. Not surprisingly, there was considerable debate within
IBM whether components was an "appropriate business" for IBM to get
3 || into, and the decision to establish the Components Division in 1961
continued to be second-guessed well into the 1960s — long after IBM
had committed itself to the point when there was no turning back.
(Case, Tr. 73515.) In short, as T. V. Learson put it in 1966: IBM
"had to become, in a very short time, the largest component manufac-
turer in the world". (PX 1900, p. 9.) If IBM were successful, the
potential benefits overrode those risks :
(a) in-house manufacture could help IBM reduce its total
costs by eliminating middle-man profits;
(b) by designing the new circuitry and the new machines
simultaneously, IBM could get the best new circuitry earliest
because IBM would not have to wait for another firm to finish its
circuit development process and make the circuit available in
1g I order to explore the circuit's potential characteristics and use
in a computer system;
17 1
13 ■
19 If
(c) unlike other manufacturers who were less integrated and
who would have to adapt generalized circuitry to their particular
;! needs, I3M would be able to enhance the price/performance of its
;j computer systems by tailoring its own circuitry to the require-
; i ments of System/360. (Case, Tr. 73245-48; see also E. Bloch,
22. ,'
;| Tr. 91563.)
" i
■{ In-house manufacture would also oermit IBM to accelerate the training j
/A s *" i
.! of computer engineers in both the characteristics and use of the new I
"i . !
.1 circuit technolocy. According to Case, it was believed that IBM "could!
:i " |
'I -289- S
synchronize the development activities between the circuit development
organizations and the computer development organizations more effec-
tively if they were in one corporation rather than if they were in
two or more corporations." (Case, Tr. 73250.)
Such synchronization was to grow increasingly more impor-
tant.. Bloch testified that as the integration level of components
increases, "more and more of a machine is on a single component.
And therefore when one has in mind the designing of a new computer
one can learn a lot by just looking at the individual components that
go into it." (Tr. 91929; see also Case, Tr. 73251-52.) As the
degree of component integration increased during the 1960s both
symbiosis in development and confidentiality became increasingly more
important reasons for in-house development.
IBM's Advanced Technology Study Committee took the long
view in 1961. It was building for the future (E. Bloch, Tr. 91929)
and considered the benefits which might be derived later on from a
long-term kind of' process worth the risk. (E. Bloch, Tr. 91928.)
;[ That long range planning paid off handsomely. Case testi-
.i
!| fied that IBM achieved the objectives that it set with respect to the
L
2
3.
4-
£
6
7
8
9-
10
11
j
12 j
13 !
14- i
15 |
16 !
17 I
IS
19 ,
;! design, development and manufacture of SLT (Tr. 73267) and that
2G !
■I
'! tne ultimate success of System/360 was "in large measure" dependent
21 ; i .
:j on the success of that circuit development activity. (Tr. 73253.)
22 : i
|| (iii) Single Family for All Applications. The SPREAD
22 !j
.; Committee recommended development of a single line of processors to
i; "meet the needs of the commercial, scientific, and communications and
I
!
:t
<\
1 -290-
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16 !
17 I
18 j
19 I
j
20 j
i
21 i
22 l
23
24
25
control markets". (DX 1404A, p. 12 (App. A to JX 38).) That objec-
tive called for a "fundamental change" in IBM's design emphasis (DX
4740: Evans, Tr. (Telex) 3925-28) but one which was thought to be
necessary for developing user requirements* At the time of the
SPREAD Report, IBM's product lines were "distinctly either commercial
or scientific in their emphasis". (DX 1404A, p. 13 (App. A to JX
38).) This was true of other vendors' product lines as well. Up to
that time, customers who wanted to do what had traditionally been
considered both "scientific applications" and "business applications"
generally acquired two computers. (Case, Tr. 73329.)
By the end of the 1950s, however, the distinctions between
business and scientific applications were beginning to blur, and
"customers themselves were not observing [the] lines between scien-
tific and business machines in actual practice". (Case, Tr. 73274-
75; see also Tr. 73276-83, see pp. 81-83, 102, 148, 162, 213-15, 239,
above.) Evans testified that "more and more" often, the "scientific
side" of a user's operation needed the data handling capabili-
ties associated with business data processors and the "business side"
needed the arithmetic and logic capabilities associated with scienti-
fic systems. (DX 4740: Evans, Tr. (Telex) 3927-28.) The history of
the 1950s and early 1960s is full of examples of "business" computers
doing "scientific" applications and vice versa. (See above, pp. 15-21
38-45, 81-83, 102, 138-49, 162-68, 206-15, 242-44.)
That user need for "dual use" was a major factor in
the SPREAD Committee's thinking. According to Evans, "One of
the premises from the beginning was there would be great
-291-
1
2
3
4
5
6
7
3
9
10
11 !
i
12 |
l
13 I
savings to the users if we could combine in the single machine the
ability to cover the full range of business applications and scien-
tific applications as well. So our concept was a single machine that
would be equally able in either of those areas". (Tr. 101052.)
Although the Committee foresaw the need for separate develop
ment of ruggedized products for military purposes , it stated that
"standard products will satisfy about 32 percent of the available
military market" and that a basic objective should be "to further
penetrate the ultra-reliable portion of the military market with the
SPREAD family". (DX 1404A, p. 44 (App. A to JX 38).)
Thus, in accordance with these recommendations, it became
i
an objective to design the NPL architecture for the "broadest possible}
range of applications . . .. equally well suited" to what had pre- i
19
20
14 ! viously been considered scientific or business computing. An
15 j instruction set and processing capabilities were to be designed to
16 j be "equally suitable to both of those classes of applications and
;i
17 jj indeed well suited to the broadest possible range of applications
ii
18 |j that one could think of" (Case, Tr. 73268-69), including process
control applications and communications control applications. (Case,
Tr. 73321). Evans testified that the name "System/360" was chosen
21 j for the new line to indicate the "full circle of the applications
il
22 jability of the machine". ( Tr. 101129.)
The combining of capability to do the whole range of applicaj-
I
tions in a single machine promised great savings to users and great
returns to IBM. It was far from clear, however, that the objective
-292-
23 i
24 !
25 !
z
3
5
5
7
a
10 ;
|
n !
12 j
13
L4
15
IS
of designing "dual purpose" computers could be accomplished without
a degradation of either performance in business applications or per-
formance in scientific applications or, indeed, in all the applica-
tion areas. Evans testified that this risk, was perceived by IBM
management and "haunted" them.* (Tr. 101052, 101129.)
"The question was whether we could build machines that in their
own right as a scientific performer would be the best and also
had the ability to do the business kind of a problem, or in so
doing would we really be building mediocrity and someone could
come along and optimize as the industry had done before and build
better scientific machines, better business data processors, and
in the process negate our plans and our aspirations." (Tr.
101052; see also Case, Tr. 73538-39.)
The risk that competitors might specialize and, in so doing,
outdistance a line of products aimed at a wide variety of applications
was compounded by the risk that, even if 360 was as powerful as more
specialized competitive machines in their specialties, customers might
reject System/360 because they just "might not see it that way".
(Case, Tr. 73538-39.) In the face of these risks, some people in IBM
: | became proponents of continuing work on the pre-existing "scientific"
17 I
l^and "business" product lines. During 1962 and 196 3 there was continued
13:!
'I a project to build a scientific computer compatible with and as a
19 ;!
;|successor to the 7094 (Brooks, Tr. 22843-44; Case, Tr. 74574); and
7Q 1
,t as late as December 1963-January 1964, a group in the General Products
:! Division led by John Haanstra opposed development of the 3 60/3 in
•1
70 :i
23
j * Case testified that IBM management "frequently" inquired of the
.360 design group whether the performance objectives for System/360
*"" Iwere beinc met for both business and scientific use. (Tr. 73539.)
,1
-293-
1 I favor of extending the 1401 line in its place. (Evans, Tr. 101187-88,
j
2 | 101275-76; Hughes, Tr.. 33970-71.)
3 | Such fears were not unfounded. As we shall see, competitors
4 j did attempt to offer more specialized systems* to meet the needs of
i
5 j certain users and were successful in competing against System/3 60
i
6 j where customers wanted such relative specialization rather than the
j
7 i more generalized range of functions which System/360 offered. Some of
i
8 j the history of the latter part of the 1960s is the history of IBM's
9 j attempts to respond to such competition.
1Q. I Despite the risks, the concept proposed by SPREAD was
11 j pursued. System/ 3 60 was designed to be a machine equally powerful in
12 | scientific and business applications and with facilities for real-
i
13 | time applications, which "machines of that age had not been able to
14 I address before System/360 with real power and versatility". (Evans,
15 i Tr. 101144.)
i
;i
16 it Weil testified that the distinction between scientific and
;j
17 ;|commercial processing was "erased" " [i]n a practical sense, with the
ii
18 -announcement of the IBM 360". (Tr. 7189; see also Beard, Tr. 10342;
19 [Friedman, Tr. 50378; O'Neill, Tr. 76194-96.)** That testimony was
20 |
21 | * The "specialization" offered by competitors was a matter of
[[degree — many "specialized" competitive offerings could be and were
22 Ijused to perform a range of applications, but were marketed as more
•1 "tailored" machines to attract certain users. For example, G.E.
23 ijinitially targeted its 600 series primarily for engineering and
;!scientific applications (Weil, Tr . 7026-27) and CDC originally de-
24 II signed its 6000 series to perform scientific applications. (Norris,
"Tr. 5617, 5618, 5629; see pp. 423-24, 672-80, 690, below.)
25
**" Weil also testified that "Since the early sixties, it really
j hasn't been economically important to design a computer system only
-294-
consistent with Weil's assessment of 360 in June 1964:
" ► . . System/360 integrates into a single set of equipment the
capability for business data processing, scientific calculation,
data communications, and process control. It seems clear that
all of these are now but facets of the basic information handling
and processing system.". (PX 320, p. 13.)
System/360 f s ability to "do the 360 degrees of the circle" resulted
in acceptance by users who could not get the same range of perfor-
I ;j mance from other architectures. (Evans, Tr. 101132-33.)* Its broad
range of applications helped simplify customers' acquisition deci-
sions, enabled them to achieve economies of scale by acquiring one
large -capacity, rather than two smaller-capacity, machines and
permitted them to reduce the required training and improve the
7 ;j efficiency of their EDP staffs. (Case, Tr. 73327-28; see also Weil,
Tr. 7059-60.) The combination of business, scientific, and other
applications in the same line also helped reduce IBM's costs. It
enabled IBM to concentrate on a single machine type with fewer sets
of program support and software and with a single set of training and
'j education for customers and IBM personnel. (Case, Tr. 7332 8-29,
7 11
•i see also Tr. 73387-89.)
3 ;l
4 Some of the benefits associated with the "erasure" of the
3 :}
'! business-scientific distinction and some of the techniques used to
\ for business or only for scientific applications , except at the
- S extreme ends of this spectrum, where you are trying to do as much
^ J scientific calculation as you possibly can within the limits of the
3 j technology" . (Tr. 7190.)
!
i
4 I
• j * The diversity of applications to which users applied System/360
■are described in more detail in the Appendix to this section.
-295-
5
6
7
3
9
10
11
12 !
13
14
15
16
1 ! effect it were also associated with the achievement of another objec-
2 ! tive of the SPREAD Committee. This was the objective of having a
3 single compatible line of processors with compatibility extending over
4- a wide performance range. Compatibility in this sense meant that
programs written for one processor in the line could be run on a
second processor, provided that the second processor had at least the
minimum memory capacity and complement of input/output and auxiliary
storage devices required by the program, and that successful execution
of the program did not depend on the speed of the CPU.* (Case, Tr.
73368-69; see also Brooks, Tr. 22681-82.)
(iv) System/360 Compatibility. The SPREAD Committee recom-
mended the development of a new family of compatible processors by
IBM:
"IBM customers' needs for general-purpose processors can be most
profitably met by a single compatible family extending from the
smallest stored-program core-memory machine to the machine for
customers growing beyond the 7 94 and 7030. There are proces-
sor needs above and below this range-it is not yet evident
that these can be compatible with the new processor family."
17 i! (DX 1404A, p. 8 (App. A to JX 38).)
18
19
20 I
i
22 I
i
23 !
24 J
25 !
The new family was to consist of at least five CPUs — those five to be
upward and downward compatible with one another. (DX 1404A, pp. 16,
25 (App. A to JX 38).) According to Evans, this concept of compati-
bility envisaged by the SPREAD Committee and implemented in System/360
was "just a mile apart from the rest of the world". (Tr. 101141.)
* These three requirements are satisfiable in 90 to 99 percent of
all the programs that normal businesses execute, according to Case.
(Tr. 73368-69.)
-296-
I ;
»- :
v •;
f
3
7
9
2
Prior to the introduction of System/360, it was generally
true that the computer lines of a particular manufacturer were not
compatible with one another. (We Ike, Tr. 19193.) Although both IBM
and a number of its competitors had achieved upward compatibility
over a "very narrow performance range" covered by two or three machines,
no one had achieved the full upward and downward compatibility over
the "very substantial" systems performance range of System/360.
(Evans, Tr. 101140-41.) Thus, several months after System/360 was
announced, Withington wrote that "the degree of upward and downward
compatibility that is achieved with System 360 . . .is certainly by
far the greatest to date". (PX 4829, pp. 17-13; see also Case, Tr.
73406-10.)
The SPREAD Committee viewed compatibility for an entire
family as a "major advance" that would appeal to customers and "sell
more processors". (DX 1404A, p. 35 (App. A to JX 38).) From the
customer's perspective, the Committee regarded compatibility as a
"powerful selling tool" because it would
(1) protect his programming investment;
; i (2) permit phased growth;
*** !
— r
j (3) minimize his investment in personnel training;
H 'i
i (4) expand the available labor market of personnel trained
!■» !
, ' u I
;| to operate in his environment;
2. ,|
j (5) simplifv the adaptation of his applications to several
! processors;
— i
(6) permit him to transfer applications among installations;
-297-
1
2
3
4
5
6
7
8
9
10
11
12
13
14
and thus
(7) provide an incentive for him to convert to System/3 60
from non-compatible families. ( Id. , pp. 35-40.)
Such benefits did, in fact, accrue to customers. For example, John
Jones, Vice-President, Management Information Services at Southern
Railway testified that compatibility was of "very great benefit" to
him as a user because
"it provides me the option of changing or upgrading the capa-
cities and capabilities of my installed network and gives me
an alternative which under some circumstances is a very
attractive one in that I do not have to do reprogramming if I
choose not to do it." (Tr. 80007-08; see also McCollister,
Tr. 11068; Friedman, Tr. 50377; Case, Tr. 73427-28; Knaplund,
Tr. 90507-08; PX 1215, p. I.)
Since System/360 was compatible over a far broader range of processor
capacities than any previous EDP line, those advantages of compati-
bility were made available "to a great many users of all sizes" — from
jj
15 ! the large, multiple-location user who would be able to reduce his
16 If training, system development and programming costs to the small first-
17 !
18 |
19 :
20 !
i
21 |
22 |
23 J
24 J
25
time user who could plan to grow rapidly without incurring reprogram-
ming costs. (Knaplund, Tr. 90507-08.) Of course, this meant that a
"great many users" would be attracted to System/360. As Brooks
testified:
"We believed the compatibility would make it possible to
make machines a lot easier to use, that it would serve the
customers better, and that it would permit IBM to furnish a
better level of customer support. ., . . [M] aking a machine more
usable makes it more marketable." (Brooks, Tr. 22692; see
also Case, Tr. 73427-28.)
Joseph Rooney, who held a position as an IBM Branch Manager
and later became the President of RCA's Data Processing Division,
-298-
1 !
2 j
4- j
5 i
i
i
I
6 j
i
7 j
i
8 !
9 !
10 j
111
testified that there was a "high degree of program compatibility"
within System/360, which provided an advantage to IBM in that
"Their clients could grow from a smaller system to a larger
system, or if the economic situations were such that they wanted
to go to a lower system,*- they could do so without having to
reinvest in their software. It also was an advantage if you had
a multi-faceted organization that had large computers and small
computers, and some commonality of applications that they wanted
to use on both types of systems. It gave the client the advan-
tage of not having to modify his software to do so". (Rooney,
Tr. 12550-51; see also Spangle, Tr. 5026; Beard, Tr. 10325.)
Withington testified that " [t]here is an advantage to a
manufacturer in standardizing on a single system set of programs
because that minimizes his total cost of development, maintenance and
customer support of such 'systems programs". (Tr. 56612.) In addition
12 | to the tremendous competitive advantage* that IBM would derive from
13 j of fering users a compatible family, the SPREAD Committee recognized
14 ; that compatibility was "clearly advantageous to [IBM's] development
15 | and manufacturing". (DX 1404A, p. 8 (App. A to JX 38).) Commonality
i
16 I in processor logic and programming were anticipated to provide IBM
i|
17 :|with economies in training of field personnel, development of program-
18
19
20
21
22
* The SPREAD Committee anticipated that this advantage was one
that competitors would not be able to overcome during the rest of the
decade unless they adopted new approaches to the achievement of
compatibility:
"Competitors appear to be relying heavily on common programming
languages to achieve compatibility. The new processor family
i| guarantees to IBM a compatibility level which will not be pos-
23 ij sible, in the 1965-1970 period for a non-compatible family of
! processors relying on common programming languages." (DX 14Q4A,
24 ij p. 40 (App. A to JX 23).)
25 ijAs we discuss later in the testimony concerning specific companies
jj(see pp. 383-84, 480-32, 619, 623, 644, 660-61, 696, 705, below), a
number of IBM's competitors did just that, albeit several years after
IBM.
-299-
L;
/^
3 !
4-1
ming and standardization of installation and maintenance procedures.
(DX 1404A, pp. 36-41 (App. A to JX 38); see also PX 1215, p. 2.)
System/360 compatibility permitted IBM to realize these and other
benefits. Case testified that training of programmers, salesmen and
systems engineers was made "considerably easier" because they had to
be trained for one group of machines instead of for different incom-
patible machines* (Tr. 73387-88.) IBM also achieved cost reductions
in manufacturing because of the ability to share parts among the
various models of System/360 and to provide common training to manu-
facturing personnel. . (Case, Tr. 73388.) Finally, IBM had to develop
fewer operating systems than it would have for incompatible processors ,
and the design of the individual models was facilitated because
commonality of design permitted the various engineering groups to
T . ; l communicate effectively and assist in one another's design efforts.
(Case, Tr. 73388-89.)
The decision to provide a compatible line over a large
performance range' was recognized within IBM as a risky one. From
/
3 ;
Q
10
U
12;
13
IS
17 |
IS •!
i
a competitive standpoint, the SPREAD Committee anticipated that a
| single compatible line could be marketed against by competitive
! salesmen who would be able to develop "knock-off s" applicable to the
i entire family. It would also provide a more nearly unitary target
i against which competitors might react more effectively with their own
22 :!
.(product and price moves. (DX 1404A, p. 40 (App. A to JX 38).) Perhaps
I most importantly, it would "encourage competition to be compatible
** A \
£- I
1 with [IBM] in order to tap [IBM's] support efforts". (DX 14 4A,
£2
-300-
1 i
2 i
i
3 !
4 j
5 i
i
I
6 |
i
7 !
8
9 !
10
11
12
13 !
p. 40 (App. A to JX 38).) That latter possibility was one that IBM
plainly foresaw throughout the 1960s and one that came to fruition in
different ways in the latter half of the 1960s and in the 1970s with
the explosive growth of leasing companies and the advent of plug-
compatible peripheral and CPU suppliers. (See pp. 750-96, 807-14,
819-26, below.)
The compatibility objective presented risks from a technical
standpoint as well. Just as the attempt to combine business, scienti-
fic and other applications in the same line raised the possibility
that the new system would do none of them as well as a more specializejd
machine, so too the attempt to achieve compatibility between very fast',
processors and relatively slower ones raised the possibility that none
!of them would be truly optimal. Case testified that
14 ;| "It was thought prior to System/3 60 that having one machine
ij architecture for both the fastest and the slowest machines in a
15 j product line and, in fact, all places in between, could not be
■j right because either the fast machines would be unnecessarily
16 II restricted in the amount of function and capability that they
jj could provide ... or alternatively, that the slowest and cheap-
17 ;j est machines would be far too expensive by virtue of having to
;j provide the richness of the instruction set that was provided by
18 ! ! the larger and more expensive machines in the product line."
ij (Tr. 73520.)
19
20
21
22
23
24
25
According to Evans, the "real challenge" of System/360 from an archi-
tectural standpoint was to build a compatible family with a perfcrmanc
range of 1 to 100 from the smallest machine in the family to the
largest — it was "something that had never been done before".* (Evans,
* The 3 60 announcement letter stated that the processors covered
a performance range of 50 to 1. Evans called this a "conservative
statement" and stated that the performance range was 100 to 1 at
announcement and had since been expanded to nearly 1,000 to 1. (Tr.
101177-78.) Evans testified that IBM successfully met its challenge
and that System/360' s "performance range, unprecedented in the indus-
trv", was a major factor in attracting customers to the 3 60. (Tr.
101144.)
-301-
L,
z
/
3
g
10
LI
12 i
14
Tr. 101057-53.) The difficulty of this undertaking was clearly
recognized by the SPREAD Committee:
"It is not evident that downward compatibility can be attained
through the whole product range. The group recommends, how-
ever, that the design requirement for downward compatibility
be stated as. a firm ground rule and that development proceed
on this basis until the Phase I review. If, at that time, it
appears that economically competitive downward compatibility
cannot be achieved across the whole processor range, then the
range shall be broken into two segments with downward com-
patibility to be achieved within each segment." (DX 1404A,
p. 17 (App. A to JX 38) .)
17 ;
Enfield, President of The Computer Software Company and
former IBM Product Administrator for the DOS operating system, testi-
fied that downward compatibility was achieved for System/360 through
the Model 25. (Tr. 19977; see also Case, Tr. 73520-24.) For IBM to
achieve that level of compatibility without incurring unacceptable
;[ expense or performance penalties at the low end of the line required a
"technological change in the way computer systems were built ... in
*~ ,1 IBM" . (Case, Tr. 73520.) That technological change was the introduc-
ia . tion of microprogramming or "firmware". ( Id . )
Microprogramming was invented by M. V. Wilkes of Cambridge
-S :, University in 1951. (JX 33, o. 5.) Case testified that IBM was the
i- '\ first computer manufacturer to use firmware in the building of computers J
.J i
£- ; j (Tr. 73222.)* Its use required the application by IBM of "new technical |
2Z 'components" (such as transformer and capacitor read-only storage) and
.i I
22.1a new design "discipline". (Case, Tr. 73521.) ]
„- 1 ' i
-* f
,'i i
2± I * That use began with the experimental SCAMP built at Hursley in
1 1960 (JX 33, p. 5) and continued with System/360. IBM continued its j
25 ! innovations in "firmware" later in the decade with the invention of
jthe floppy disk. (Case, Tr. 73223.) I
I :
! -302- i
1 I
2J
3!
I
i
I
4 |
5 j
6 i
i
7 i
i
8 i
s|
I
10 I
I
ii !
i
12 |
i
13 !
i
14 |
15 i
i
i
16 i
Through the use of firmware (rather than hardware or soft-
ware) IBM was able to achieve a number of the design trade-offs which
System/360 required.* It was the "technical device . . . most
responsible" for the fact that IBM System/360 computers were able to
be designed efficiently for both business and scientific applications
(Case, Tr. 73225; see also Evans, Tr. 101142-43) , as well as the
method by which IBM was able to achieve full upward and downward
compatibility. **
Some measure of the success that IBM achieved in imple-
menting the architectural objectives laid down for System/3 60 may be
gleaned from the longevity of that architecture. Compatibility and
applicability to a wide range of applications were characteristics
(assuming that they were effectively implemented) that would undoubt-
edly be desirable in future systems. Accordingly, Case testified:
"We tried to develop the computer architecture which would
be extendable, which would be useful not only for the machines
that were going to be announced in 1964, but also for subsequent
l! machines as far into the future as we could plan for. ... We
*' :! were thinking in terms of 15-20 years . . . and we would like to
j have had that last even longer if that were possible." (Tr.
18 j 73347.)/
19 ''
20
21 j
22
23
24
25
* The need for such trade-offs was understood by the SPREAD Com-
mittee, which imposed as an "engineering ground rule" the use of
| microprogramming controls unless "conventional" control systems could
jiprovide a cost/performance improvement of better than one-third. (DX
|!l404A, p. 20 (App. A to JX 38).) Microprogramming was used in the
i!system/360 processor models 2020, 2030, 2040, 2050, 2065, 2067 and
"2085. It was not used in the 2044, 2075, 2091, 2095 and 2195. (JX
38, <f 8, pp. 5, 6.)
** As we explain later, it was also the means by which System/3 6
enabled users to run programs written for earlier IBM computer systems ,
|| r To lengthen the life of the 360 architecture, the 360 design
jjcrouD chose a memory addressina structure that provided for the eventual
II -303-
L i The architecture of System/360 lasted through the 370 into the 303X
Z\ and 43XX lines and continues to the present time. (Evans, Tr. 101133;
S || see also H. Brown, Tr. 82972; PX 4505, p. 1; PX 4531, p. 1; DX 860,
4-i p. A; DX 9405, pp. 552, 1013.)
5- ji (v) Emulation. Withington called IBM's introduction of
r
g- the System/360 "a substantial risk" for two main reasons:
"One, IBM adopted a new machine architecture and a dependence on
systems programs to cause the machine to be usable to the users.
This was a large step in terms of the evolution of machine
architecture and design, and it was not immediately certain
either whether it would work well or whether the users would
accept it.
"The second primary area of risk was the lack of compati-
bility between the 360s as announced and the predecessor IBM
machines .
"It was immediately obvious that the willingness of the
customers to reprogram from the older machines to the 360s was a
major question relating to its probable degree of success."
(Tr. 56592-93.)
The disadvantage of offering a new incompatible line was
clearly recognized by the SPREAD Committee. It was, however, a
disadvantage that had to be overcome rather than avoided if the
Committee's concept for the new line was to be instituted. As the
7
3
9
10 ;
ii
iz
13
1+
15
16
17
si
;l attachment of 16 million bytes of main memory without modification and
2G i about 2 billion bytes with only a "small" modification. That eliminated
jone of the "major reasons" that previous architectures had been short-
ZL ;] lived: the limitation on the amount of main memory that could be effec-
tively used with those architectures. (Case, Tr. 73347-4 9.) The 8-bit
22 [byte was another factor which gave System/360 architecture greater
j longevity than previous systems. It permitted the use of 36 in appli-
22 : cations that required character sets which made those applications
[difficult to achieve on the 6-bit byte and 7-bit byte computers which
2d '{preceded System/360. (Case, Tr. 73349-50.)
-304-
SPREAD Report noted, "Since [the new] processors must have capabili-
ties not now present in any IBM processor product, the new family of
products will not be compatible with our existing processors." (DX
•J 1404A, p. 12 (App. A to JX 38)', emphasis in original.)
The SPREAD Committee anticipated that the new capabilities
provided by System/360 would induce many users to switch to System/360
despite the need to convert their programming. Indeed, for many of
these users, the very fact that they wanted to implement new functions
rendered the entire question of conversion moot:
"... While incompatibilities are a marketing disadvantage,
it should be noted that systems reprogramming will, in many
cases, be required, independent of the processor used. This will
occur whenever the user wishes to obtain the benefits of any of
the following:
;j
|{ "a. Random access rather than batch processing
"b. The integration of communication facilities
"c. The simultaneous operation of multiple processors
"d. Multiprogramming to achieve efficient on-line operation."
( Id. , p. 12; see also Currie, Tr. 15184-85; Withington, Tr.
57683-84.)
In "many cases", therefore, the reprogramming effort involved
1
j in switching to System/360 would be no more than a "natural outgrowth"
i j of the systems improvements that the user wished to achieve--improve-
► jments that would require a new programming effort whether or not that
•i
L .1 user switched to an incompatible processor. (DX 1404A, p. 12 (App. A
> ito JX 38).) However, the Committee also recognized that " [sjome
I. 'customers [would] be dissatisfied unless an alternative [was] provided
■j
• jto permit utilization of [their] prior machine investment". (Id. ,
-305-
L
2
3
4-1
si
I
6- j
T\
3 ;
3 ;
10 I
11 l
12 I
i
13 |
14- :
p. 39.) IBM provided customers with that alternative in the form of
emulators . *
Other manufacturers of computer systems also recognized the
desirability of facilitating conversion and provided users a number of
aids, such as simulators**' and translators, 7 ^ to ease the transition
between incompatible systems.^ As late as August 1963, IBM was still
working on software simulation as a means of providing System/3 60
compatibility with prior systems. However, work on providing con-
version through emulation had commenced within IBM prior to that time.
(Hughes, Tr. 34047-48.) On August 1, 1963, D. H. Furth, Corporate
Director of Programming, sent a memorandum to Evans expressing the
view that it was "feasible" to use read only memory control (micro-
programming) to achieve compatibility. He wrote:
* An emulator is a combination of hardware and software that
^* j permits one computer system to execute programs written for another
'! system. (JX 1, p. 45.)
;| ** A simulator performs the same function as an emulator, but is
17 !| implemented entirely in software. (Goetz, Tr. 17654.)
■»
IS ;j /A translator is a computer program that takes as input the source
:| programs of a particular computer and translates them as closely as
IS ■} possible to an equal program in the same or a different language that
■\ would run on the equipment to which conversion is desired. (King, Tr.
20 1 14769-70.)
i
'i
21 I • // GE offered a 1401 simulator which permitted programs written for
] a 1401 to be run on its 400 line and a 7090 simulator which permitted
22 :! programs written for the 7090 or 7094 to* be run on GE ' s 600 line.
j (Weil, Tr. 7029-32.) RCA developed a simulator that allowed programs
23 -written for IBM's 650 computer to run unchanged on the RCA 301. (DX
; 561, p. 13.) Honeywell offered a LIBERATOR program which translated
2- j IBM 1400 Series programs into programs usable on the Honeywell 200.
■I (R. Bloch, Tr. 7578, 7588-39, 7605-06, 7886-39; Goetz, Tr. 17652-54,
25 'j 13822-23; Enfield, Tr. 20052-54; DX 6661, p. 6.)
!
;! -306-
L
I
I
Li
"Since such a hardware simulation would appear to be very
economical from the customer's point of view and since it would
eliminate some half dozen simulators from an already mountainous
Programming Systems load, it would appear reasonable to pursue
the realization of this feasibility as part of the overall NPL
program." (DX 2872.)
3
!!
T I
3 J
i
0;
* i
By October that recommendation had been accepted, and Brooks wrote
, that "We are hopeful that microprogrammed simulation can add substan-
;| tially to the bag of tools for aiding conversion". (DX 2900.)
During 1964, IBM announced microprogram-based compatibility
features on System/360 for the 1401, 1410, 1440, 1460, 1620, 709,
7010, 7040, 7044, 7070, 7074, 7080, 7090, 7094, and 7094 II proces-
sors.* (JX 38, pp. 30, 289, 292, 334, 526; DX 14305.) Withington
testified that System/360 was "the first major use of microprogramming
for purposes of establishing backward compatibility."** (Tr. 56606.)
He also testified that
3 '
£• : "implementation of emulation using control store and micropro-
grams, while it is more expensive [than software emulation], is
regarded by users as preferable in most cases because it is so
|! much faster". (Tr. 56371-72; see also DX 2900.)
■6 :!
•j The provision of emulators on Svstem/360 afforded users a
| hardware alternative to conversion. (PX 449, p. 9.) It permitted
3
; them to transfer jobs to System/360 and to concentrate on new application
.3 jj j
:j areas without immediately having to convert their existing applications, j
a; j i
i
!j * Case estimated IBM's cost of developing the 1401 compatibility
£; feature on the Model 30 as $200 thousand and the cost of developing
: jthe 7090 emulator on the Model 65 as one-half million dollars. (Tr.
n ; 74557-62.)
- i
i± ■■} ** Withington defined "backward compatibility" as "the use of
(emulation . . . for the purpose of allowing programs written for a
jc jmanufacturer ' s prior computers with different instruction sets to be
~* '.executed on the newer computers". (Tr. 56606.)
>j
:! -307-
if
(JX 38, p. 30.)
Although programs run in emulation generally ran slower
than they would have if rewritten to run in native mode on the new
systems,* they could be run effectively enough to permit users to
forego reprogramming if they chose to do so. (Beard, Tr. 9057-58,
c ![ 9956-57, 10029-30, 10313-19; see also R. Bloch, Tr. 7608-09, 7614-15,
7 ,j 7881-82; McCollister, Tr. 11287-89; Rooney, Tr. 11853, 12395-96.)
Goetz testified that emulators were generally considered an "effective
means of running programs from one computer system on another".
(Tr. 17655, 13778.)
3ecause 360 was incompatible with IBM's second generation
equipment the conversion from IBM's second generation equipment to
360 involved as large a task for users as would converting to
.' another vendor's systems. (Beard, Tr. 9058-59, 9953-60, 10324-25;
McCollister, Tr. 11069; Goetz, Tr. 18935-36; Enfield, Tr. 200-20-21.)
Indeed, in some instances conversion to non-IBM equipment would have
3
9-
Li
12!
n
15
is it
;i been easier than conversion to 360. Weil testified that GE was
17 3
!{. initially "overjoyed" with the announcement of System/360 because GE
IS ;[
had introduced a system "designed to displace" IBM's 7090s and
is i
r 7094s and believed that "it would be easier ... to convert from
20 :i
1 the 7090/7094 to the 600 series" than to 360. (Tr. 7060-61.)
22 *
I * Of course, such programs might very well run faster in emulation
22 mode than they had in native mode on the equipment for which they were
; written. For example, Enfield testified that a 360/30 operating in
2± 1 emulation mode could execute 1401 programs 3 to 3 1/2 times faster
•| than a 1401. (Tr. 20263.)
25 \
! -308-
Jones testified that Southern Railway ran benchmarks which showed
that conversion from an IBM 7000 Series system to an IBM 360 was
"about equal in difficulty" to conversion to an RCA or Burroughs
machine, but not as easy as conversion to a Univac 1108. (Tr. 79042-
43; see also Hart, Tr. 81936.)
Nevertheless, IBM was successful in getting users to
convert to System/360 from IBM second generation systems. (Withington,
Tr. 57680-81.)* One reason for that success was, undoubtedly, the
benefits that users were able to derive from System/360 's improved
price/performance and new capabilities. As Withington agreed,
"if [users] perceive it to be in their economic interest,
[they] will absorb the cost of conversion for the future benefit
that they expect to receive from [a] newly acquired computer
system". (Tr. 57677; see also Hart, Tr. 80222-24.)
Hart, head of the Computer Science Department of the General
Motors Research Laboratories, testified that his department went from
a 701 to a 704 to a 7090/94 to a System/360. (Tr. 81938-39.) Several
;! years after these changes, Hart wrote "conversion costs must be taken
* || into account when changing computers; however, in retrospect, the
3 ! value of each of the above changes far exceeded the costs incurred" .
* ■j (DX 3753 (Tr. 80193).) He explained that improvements in sheer computer
2 ! speed, reduced computation costs, and the availability of "new kinds
i
I j of capabilities" were all reasons for changing computer systems. To
i
3 ; * It should be noted that IBM's success in getting users to convert
| was not unique. According to Withington, between 1964 and 1970, some
i. 1 90% of second generation equipment users converted to a non-compati-
| ble computer system of either the same or a different manufacturer.
s '! (Tr. 57677-83.)
-309-
Lj
2 J
i
i
3 !
3
0+
7 |
3
9
10
II !
decide whether conversion is justified, "you take into account the
costs of making the change, the benefits which are going to result
from the change, [and] determine whether the benefits exceed the
costs." (Tr. 80222-24.)
A similar cost/benefit analysis was performed by NASA,
circa 1965. NASA had just made a "large purchase" of second genera-
tion machines to lower its operating costs, when a "new series of
equipments"* became available with multiprogramming capabilities, I/O
flexibility, memory sizes, program logic and the ability to use
remote I/O devices that made it
"possible to effect a consolidation of [NASA's] ADP resources
. . . into a powerful central complex without compromising
. availability, quality or power available to any user. At the
-^ jj same time the cost per computation of these newer machines was
T - 1 considerably lower than their old second generation equiva-
^ :! lents". (DX 5440, pp. 2-3.)
^; NASA decided to convert "at the earliest possible time". ( Id. )
NASA's analysis of the conversion difficulties was:
U3
IS ;{ "This conversion has created a considerable workload and has
resulted in overlapping of older and newer equipments with its
attendant increased rental costs during the conversion period."
(DX 5440, p. 9.)
: | It concluded, however,
17
IS
IS]
] "The benefits from the more complex software and the flexibility
20 i of the new machines far outweigh any conversion cost we mav
| incur." ( Id. )
21 1
ij Despite the powerful incentives that users had to incor-
jporate System/360 's new capabilities, it seems clear that 360 would
22
24 1
.j * The new equipments included IBM 3 60s, Univac , CDC and GE computers.
?c ! ! (DX 5440, p. 5. )
— -|
i
,i
I
•! -310-
have been far less successful without emulators. Xerox's Competitive
Reference manual noted the success of IBM's emulation approach to
converting second generation users to 360 (PX 449, p. 9), and
McCollister testified that it was a "very widespread practice" in the
late 1960s for IBM users to choose the option of emulation on 360..
(Tr. 11287-38.) An IBM Corporate Programming Study based on a
: November 1967 customer survey estimated that "more than half of the
systems hours now being used by our Models 30, 40, 50 and 65 are being
used in emulator mode". (PX 2161, p. 3.)
(vi) System/ 3 60 Software. Prior to the advent of operating
systems, each programmer had to write instructions that would schedule
his tasks and control the various equipments he required for his
particular jobs. As computer systems became faster and more complex,
it became increasingly important to manage efficiently the resources
they provided. Operating system software relieved programmers of the
, need to incorporate scheduling instructions in each program they
; | wrote and, -in effect, turned over the job of scheduling to the computer
J '\
,1 itself. According to Dr. Perlis, operating systems enabled users to
;i
'} "take advantage" of a computer's total processing power, including
9 \
i
i i
i
M
_ i
i
r i
V
I
T
m
;l
£ i
■j
2 1
1
its multiprogramming and multiprocessing capabilities . (Tr. 1848-49;
see also Welke, Tr. 17113; Goetz, Tr. 17476-77; Enfield, Tr. 20737- .
38; Case, Tr. 73443.)
Given the complex "new market demands" and modes of use at
! which System/360 was being aimed — i.e., "multi-terminal, on-line,
IS
I real-time, multiprogramming operation" (DX 1404A, pp. 7, 8, 9, 54
I (App. A to JX 38)) — it was imperative that IBM automate as much as
-311-
L ; possible the system's resource management task. IBM embarked on the
2 ; creation of a set of operating systems of varying complexity.* The
3 I most complex of these, OS/360, was particularly ambitious.
4- i OS/360 was designed to let customers "make the maximum
'■. possible use of the relatively greater speed of the . . . System/360
5* central processing units". (Case, Tr. 73438.) Since multiprogram-
ming was anticipated to be a "normal" mode of use, facilities (such
as an interruption mechanism) were to be included to make multi-
programming "easier, straight forward and efficient". (Case, Tr.
73438-39.) In addition, OS/360 was to contain facilities that would
permit programmers to develop applications more efficiently, optimize
the utilization of peripherals and simplify maintenance. (Case, Tr.
; 73438-41.)
Within IBM, it was recognized that "no one [had] ever
undertaken a programming task of [OS/360's] magnitude". (PX 1092, p.
4; PX 1900, p. 8.) Dr. Perils called OS/360 a "really major effort",
id ;f
;j one which "generalized every aspect of operating systems known at the
* 'J
' time and tried to in a sense build a system that would be all things
' to all men". (Tr. 1887.) Mr. Welke, President of International
IS !j
'[Computer Programs, called OS/360 "a major programming effort" which
2Q 1
.ranked "along with . . . the creat undertakings of mankind". (Tr.
21 j .
■\ 17313; see also Rooney, Tr. 12576.)
22 ;i
7;
3 !|
9
10.
LI !
12
12
14
j * To account for the varying degrees of speed and complexity of
2i ] operation that users might desire, I3M provided with 360 a "spectrum
[of operating systems . . . each of which offered a different memory/
25 j function trade off for the customer". (Brooks, Tr. 22759.)
•i -312-
•i
r
,1
' .1
So ambitious an undertaking entailed significant risk, and
as we shall see, OS/360 was quite costly and difficult to perfect.
Apart from the difficulty of constructing the operating system at
If all, there was the additional .risk that users would reject the multi-
' programming environment — an environment that was most often not used
in earlier generation systems. (Case, Tr. 73526.) That would mean
that IBM's investment in the hardware and software needed to permit
multiprogramming would be reflected in System/360' s prices and would
have accomplished little more than to make the systems less competi-
tive. In addition, OS/360 1 s "extensive" resource management, data
management, languages, aids to program development and error recovery
techniques did not come "without a price". (Case, Tr. 73527-29.)
The use of those capabilities would take up auxiliary storage space,
main memory space and time on the CPU — an "operating system over-
head" . (Case, Tr. 73529.) There was a significant risk that users
would be unwilling to accept such "overhead" for the richness of
: | function provided by OS/360. (Case, Tr. 73528-30.)
i
I OS/360 did, in fact, run into "difficulties in design, in
H
•(correctness [and] in completion".* (Perlis, Tr. 1887.) However,
n i
:| "when the system finally worked it had properties that were beyond I
3 i !
; about any other operating system around". ( Id. ; see also Palevsky, !
M • !
.jTr. 3180; Rooney, Tr. 12576; Currie, Tr. 15186; Welke, Tr. 17308-12.) j
2 :| I
jit must be remembered that OS/3 60 was only one of five general program- I
4. 1
| * Many other computer systems suppliers experienced similar diffi-
5 jculties in designing complex operating systems during the 1960s.
i(See discussion below, pp. 364-66, 479, 502, 568-72.)
-313-
L
2
Z
4-
3"
S A
8 It
a
la
ii
12
13
14-
15
17
IS
-i
it
20
i
ft .!
ming packages that IBM announced in 1964 for use with System/360.
(Brooks, Tr. 22759; McCarter, Tr. 88388; JX 38, \\ 9, p. 6.) The
others — Basic Programming Support (BPS) , Basic Operating System
(BOS) , Disk Operating System (DOS) and Tape Operating System (TOS)
were less complex sets of systems software. These operating systems
"worked reasonably well from the start" and were well accepted by
j ;j customers. (Withington, Tr. 58596-600; Enfield, Tr. 20947-52, 21120;
Brooks, Tr. 22853-54, 22862-63; McCarter, Tr. 88388; DX 1410; PX 6217,
pp. 3-4.) DOS in particular, which was less complex than OS/360 but
still 25 to 50 times as complex as the systems software provided with
the 1401, was highly rated by users and widely used. (Enfield, Tr.
20299-300, 20741-42, 20088-89, 20943-48.) Case testified that "if it
had not been for the operating systems for System/360 . . . the value
of that equipment to users would have been considerably less than it
was and . . . the orders and acceptance for that equipment would have
been a lot less than they otherwise were". (Tr. 73443-44.)
(vii) System/360 Peripherals. Case testified that one of
the design objectives for System/3 60 was to provide "a wide variety
of peripheral equipment that could be combined in a very wide range I
of configurations". (Tr. 73416.) Prior to announcement, the "breadth" j
j
!of 360 's peripherals were viewed within IBM as a prime motivation for !
I . |
:j users to re-systemize their applications and convert to 360. Thus, I
tin January 1964, Brooks wrote: ' j
2j i !
| "Even though present applications can be simply mapped onto
2.~\ System/360, many new system concepts will offer substantial j
j incentive for the customer to re-plan his application. These
2s 1 include file orientation, communication facilities, large memories, \
~~* bulk stores, etc." (DX 1172, p. 1.) j
.1 :
I ;
•j -314- I
The April 7, 1964, 360 announcement contained "many features
different from those previously offered by IBM" . Included in the
announcement were "direct access storage devices (including the 2311
disk drive, the 2321 data cell and the 2301 drum storage device);
; control units, high performance tape drives (including the 2400 series
s ! and the 7340 Hypertape drive Model 3) ; visual display units (includ-
ing the 2250) ; 7770/7772 audio response units; communication and data
acquisition equipment (including the 1070 process communication
system); and a printer, the 1403-N1". (JX 38, \ 6, p. 4.) IBM also
announced numerous additional peripheral devices for use with System/
360 subsequent to the April 7 announcement — including the 2314 disk
drive, new terminals, additional models of the 2400 tape drive, the
2420 tape drives and optical character recognition equipment. (Id. )
3 i
9-
o ;
x
2,
i:
The 360 announcement letters describe some of these peri-
pherals as follows:
|j 1015 Inquiry Display Terminal: "Used to interrogate and
£ if
j| receive visually displayed replies from a System/360, mdl 30, 40 or
.3
|| 50." (JX 38, p. 43.)
\ 1070 Process Communication Svstem: "A Tele-processing System
L9 i[ |
:} designed for two-way data communication between remote process loca- i
zo:j j
:[ tions and a central data processing area." Applications include "control i
21 1
Ij of oil fields, petroleum and natural gas pipelines, utility' distri- j
: bution systems; data collection in refineries, chemical plants, steel !
a | !
[mills, and manufacturing processes .... The 1C70 forms a complete !
2A i ■
. tele-processing system when attached to . . . System/360, via a 2701 ;
-315-
Data Adapter or 2702 Transmission Control". ( Id. , p. 39.)
1403 Model Nl Printer (originally announced as 2201 Mod. 3) :
"[plrinted output for a System/360, model 30, 40 and 50. . . . maximum
speed, 1,100 1pm". (IcL/- pp. 84, 198.)
1418 Optical Character Reader; "Optically reads data from
printer card or paper documents. ..." ( Id. , p. 70.)
1419 Magnetic Character Reader: "Reads magnetically
inscribed data from card and paper documents. . . . Documents read at
maximum rate of 1600 documents a minute." ( Id. , p. .71.)
2250 Display Unit: "A cathode ray tube unit for displaying
output in alphameric and graphic form for System/360 An
input/output unit which offers increased speed and flexibility for
; file inquiry, inventory control and dynamic monitoring of computer
operations and continuous process control." ( Id. , p. 85.)
2301 Drum Storage: "High performance random access storage
for a System/360, mdl 50, 60, 62 or 70. . . . [D]esigned for
applications such- as main memory extension, programming system resi-
dence and table or index storage." (JX 38, p. 86.)
2311 Disk Storage Drive: "For fast, flexible access . . .
85 millisecond average access speed . . . 156 KC/312 KD data rate . . .
adapters enable the 2701 to communicate . . . with the following
terminals :
"1060 Data Communication System ....
"1050 Data Communications System ....
"1033 Printer .....
"1031 Input Station ....
"1070 Process Communication System ....
"1053 Printer ....
"AT&T 83B2 Selective Calling Terminals ....
"Western Union Plan 115A Outstations ....
"Common Carrier TWX Stations ....
"1009 Data Transmission Units, 1013 Card Transmission
' j" Terminals, 7702 Magnetic Tape Transmission Terminals or 7740
Communication Control Systems ....
"7701 Magnetic Tape Transmission Terminals or 7750
Programmed Transmission Control Units ....
"7710 Data Communication Units, 7711 Data Communication
Units, or another System/360. ..." ( Id. , p. 90.)
2702 Transmission Control: "For on-line attachment of
i
various asynchronous input/output devices via private or commercial
j common carrier transmission facilities to a System/360. . . . [A]
;l modular unit with a variety of features to meet a customer's data
f .!
■»
i communication needs with a System/360". (Id. , p. 9 3.)
Multiplexor Channel: " [P]ermits simultaneous operation of
I/O units on time-sharing principle . . . primarily designed to
>
[handle multiple terminals and low speed I/O units." ( Id. , p. 31.)
L i
J The combination of those and other peripheral product
-317-
1 i
i
2 i
3 j
5 I
« ;
7 |
i
3 |
a !
LI I
12. i
i
13 j
is i!
announcements and the announcement of six central processing units
with a wide range of memory options was "unprecedented in the
industry". (Evans, Tr. 101134; JX 38, pp. 14-25; see also PX
4829, pp. 16-18.)* This range of peripherals was important to
customers when considering System/360 against competitive systems
because it greatly expanded their ability to change or add to their
systems as their requirements changed and "played a large part" in
customer decisions to go to 360.** (Evans, Tr. 101134; see also PX
4829, p. 13; Withington, Tr. 56770-71.)
The broad range of peripherals announced with 360 promoted
two of the SPREAD Committee's primary objectives — the creation of a
single system able to perform all applications and one that would
address increasingly important new applications (i.e., multi-terminal,
on-line, real time applications). The announcement of new disk drives,
tape drives, communication controllers, card and printer I/O, ter-
«i!
1 * " [T]here has never been a time when any of the general purpose
^' <l • competitors to IBM have offered more variations on peripheral equip-
ment, the total breadth of applications and systems program functions
and the total number of alternative processors" being offered by
j IBM. (Withington, Tr. 56770.)
19 i ■
; | ** We do not mean to imply here that all of the peripherals announced i
-S .j with 360 were successful. A number were soundly rejected by users. j
1 For example, the 2321 data cell was a "major product failure" which j
2i |! failed to achieve success because of unreliability; IBM had to super- j
j sede the 1015 terminal with the improved 2260 because the 1015 was j
22 j simply not competitive; and Hypertape turned out to be a "failure" even j
] though it was judged within IBM to be technically superior to com- j
23 [ petitive offerings. (Case, Tr. 74205-06, 72787-88; Withington, Tr. j
1 58534, 56475-76; JX 33, pp. 346-47; PX 6671, pp. 15, 26; PX 2990, p. R3 ti
24. | DX 13949.) As we discuss below, IBM acted quickly to shore up areas j
;j in the product line which were not judged to be superior to competitive .
25 j offerings. (See below, pp. 390-95.) !
I !
t ;
j -318- \
L j minals, audio response equipment, magnetic and optical character
2. : | readers and paper tape and process control units meant- that users
3. : could build configurations specifically tailored to their application
i -
4.|- requirements — whatever those requirements. Dr. Gibson testified that
; one of the features of 360 that permitted it to be used for both
scientific and business applications and "erase the previous distinc-
j \ tion" was "the very wide range of input/output equipment easily
3
9
i
Z
attachable through a common interface, . . . [which] made it relatively
simple to configure a commercial system ... or one. optimized for
scientific computing". (Tr* 2948-49; see also JX 38, p. 28; PX 3638,
p. 1; PX 4829, p. 13.)
In addition, the variety of remote I/O and communications
equipment offered with System/360 underscored 360 's emphasis on new
applications. Weil of GE wrote that System/360 "has major strength
in a variety of new mass storage devices and a whole new array of
; remote terminal equipment .... It has many of the features which will
: [ make possible its application in direct access systems." (PX 320, p.
■n
'(13.) Displays, remote data collection equipment, remote process
'{control equipment, communications controllers, data communications
' equipment and on-line banking equipment were all made available to
20 !i
'{permit users to bring the power of 360 to bear at the point of trans-.
11 ]
.j action — in real time. The ability of a System/360 to communicate
22 !
[with other computers or terminals "opened ua a whole new gamut of
a i '
j applications in industries, airline reservations industries, modern
i- ;
■j business, so that remote stations could have access to the enormous
?s i
1 -319-
L
2
3
5
data in a central computer and do so in real time". (Evans, Tr.
101136.) The ability to do such applications resulted in sales of
systems that otherwise would not have been sold. (Evans, Tr. 101135.)
The importance of System/360' s peripherals to the success
.of the product line cannot be overestimated. As Mr. Norris of CDC
testified, the speed, performance and price of peripherals are "impor-
tant considerations in determining to acquire one system or another" ..
(Tr. 6019-20; see Withington, Tr. 56239, 56246-47.) Thus, even a
single peripheral device— such as a disk drive, terminal or printer —
which is sufficiently better than competitive offerings can swing the
total system decision. ( Id . ; Currie, Tr. 15495-96; Rooney, Tr.
12048-49; DX 13949.) In this respect, of all the peripherals offered
with System/360, the 1403 Nl printer and the 2311 and 2314 disk
drives were most critical to 360' s success.
1403 Nl Printer. We discussed earlier the importance of
the 1403 printer to the success of IBM's 1401 computer system, and
how that printer gave IBM a "tremendous advantage" in the marketing
of systems until competitors began to offer "satisfactory alternatives"
7
3
ia ;
u
12.
13
IA
15
*6 ,
IS ;
ijby 1963 or 1964. (See above, p. 143.) In 1964, IBM announced
13
:|the 1403 Nl Printer for use with System/360. The 1403 Nl ran at
20 J.
[almost twice the speed of its predecessor (1100 lines per minute
21 1 .
:j compared to 600 for the 1403) and cost only about 15% more than the
22 l \
[1403. (Evans, Tr. 101137; DX 3617; see also Enfield, Tr. 20266; JX
j38, p. 207; DX 573, pp. 4, 6.) At the time of its introduction IBM's
<■■* i
.{competitors did not offer a printer that matched the 1403 Nl in print
25 i
J ' -320-
quality, price and speed. (Evans, Tr. 101137; see also Case, Tr.
72881.) IBM's competitors recognized and acknowledged the excellence
of IBM's printers . Beard (former Chief Engineer of RCA's Computer
| System Division) testified that RCA began offering the 1403 with its
Spectra Series because there were applications for which customers
desired print quality "of a very high standard" . Such customers
"insisted" on "1403 chain printer type quality" and "after resisting
these requests some period of time" RCA acquiesced and "put the 1403
into the RCA computer line". (Tr. 10322-23.)
The 1403 Nl was particularly important to System/360 's
ability to perform certain business applications. For a customer with
applications such as payroll, billing, accounts receivable and inven-
tory control, the ability of a computer system to do his work is
determined "in large measure" by the speed, quality and reliability
of the printer. (Evans, Tr. 101137; see also Currie, Tr. 14971-72;
; Withington, Tr. 56253.)
Currie testified that XDS was at a "disadvantage" to IBM
with respect to its line printer for customers that wanted to do "any
.! significant amount of business data processing". (Tr. 15459.) As
'j late as 1969, XDS was only "marginally competitive" in peripherals and
► 'I 8
i i j
1 its line printers "were not acceptable to some of our users". Those . I
j printers lacked the range of "speed/performance" that some customers |
I 1 I
| wanted and did not produce as high a quality print as a chain printer j
i t i
lor a train printer. (Currie, Tr. 15006-08.) !
' !
;! CDC also experienced "substantial problems" in marketing some
-321-
t
2
3.
4-
S
T\
a |
si
i
i
u!
12
1* i
i
i
of its computer systems because they incorporated printers that "lacked
sufficient reliability to meet normal customer expectations" and had
"a poor print quality, in terms of wavy print". To help solve these
problems CDC acquired the Printer Division of Holly Carburetor in 1966.
(G. Brown, Tr. 51528-29.) CDC ultimately developed a "1403N-1 type""
printer of its own, but it had to be "reworked and re-developed" in.
the 196 9-70 time frame in order to effect reliability improvements.
The changes resulted in a design that was "more like the original IBM
design". (G. Brown, Tr. 51541-46.)
While CDC attempted to copy the 140 3 Nl design and RCA simply
incorporated it into RCA's product line, Grumman Data Systems took
advantage of the 1403 Nl's superiority by offering to attach it to a
number of non-IBM computer systems. As late as 1975, an advertisement
for Grumman Data systems stated:
ia
"For years people have been trying to imitate the IBM
1403. Unsuccessfully. Now, with the Grumman Printer Controller
you can connect your present computer to an IBM 1403 and give
yourself the best printing in the business.
17 ;[ "The IBM 1403 has built an extraordinary record. Highly
|| reliable, high speed operation. Unusually consistent, clearly
IS jl readable printouts. (No wavy lines so typical of drum printers.)
i! Type fonts your operator can readily interchange. And, of course,
19 ;i it handles form changes easily.
25 JGrumman later offered the 1403 Nl for attachment to Burroughs, Data
20 I "With the Grumman Printer Controller you can improve your i
| printing quality, speed, and reliability. All at an attractive,. j
21 -| and perhaps, money-saving price. Speaking of price, you can buy i
• our controller or rent it. We provide maintenance of course. !
22 [ t . !
"With our printer controller you can connect the IBM 1403 |
to your present DEC, Xerox, GE , or CDC computer. We'd like to j
I hear from Burroughs, Univac and the other computer users, too." !
24 , | (DX 94B.) j
-322-
General, Digital Scientific and Univac computers. (DX 2782A; DX 7984.)
The 140 3 Nl was also offered with Computer Machinery Corporation
computers. (DX 11665.)
Gordon Brown testified that the quality and reliability of a
■. printer is "an extremely important criterion in the selection of a
computer system". (Tr. 51528-29.) The 1403 Nl was a real boon to
the acceptance of 360.
System/360 Disk Drives. As we discussed earlier, IBM's
superiority in direct access storage technology during the 1950s and
early 1960s contributed greatly to the success of IBM's first and
second generation systems. (See above, pp. 91-95,149-53.) IBM maintained
that superiority with the disk drives introduced for use with System/
360. Both the 2311 and 2314 were substantial improvements over IBM's
earlier disk drives and both proved critically important to the
success of System/360. These disk drives were more than just superior
to competitive offerings, they were unique in the industry: there
simply were no similar competitive offerings for several years after
their introduction. Thus, they gave IBM a competitive advantage in
the marketing of 360 systems that competitors were unable to match
until the late 1960s, and even then, competitors were able to do so
I J
I only by adopting, in one way or another, IBM's disk technology.
:!
■\ IBM announced the Model 2311 disk drive on April 7, 196 4.
I ■;;
;| The 2311 had approximately twice the access speed, twice the data
• i
irate and two and one-half times the storage capacity of the 1311.
i. 1
|(Case, Tr. 72741-42; JX 38, p. 86; PX 4252, p. 1; DX 3554D; see also
I -323-
•i
I
z
3
4.
3
e
71
3
9-
IQ
IZ!
Is !
14
15
IS ;i
17 J
IS
: earlier model". (Tr. 9597.) Withington agreed that both the 2311
19 i -
itand 2314 were unmatched by comparable competitive products during the
2d !
{initial years in which they were marketed. (Tr. 58800, 56240-41.)
21 i
:| IBM foresaw and depended upon the widespread acceptance of
22 :, »
idisk. drives as a key factor in the ultimate success of System/360.
>•«» .'
cz ;
I IBM Vice President Knaplund testified:
24 I
j "An important element of the System 3 60 forecast was the antici-
ys. ' pation that. disk files would be used extensively, both in applica-
-324-
Enfield, Tr. 20264-65; Haughton, Tr. 94998.)
IBM announced the 2314 disk drive on April 22, 1965. The
2314 had a faster access speed, double the data rate and almost four
times the storage capacity per spindle of the 2311. (Case, Tr.
.72742-43; JX 38, pp. 86, 439; DX 3554D; see also Haughton, Tr. 94998.)
Beard testified that the 2311 represented a "technological
advance" over prior random access storage methods. "It provided not
only . . . fast access time but it provided . . . for the first time, the
degree of reliability that was required of random access devices
.... [I]t was really the first very reliable disk file that . . .
was offered by anyone". (Tr. 9048-49.) Beard also called the 2314
an "advance over prior random access devices", adding that his comments
on the 2311 applied "perhaps more importantly" to the 2314 because
the 2314 offered greater storage capacity and a more "practical cost"
for random access storage than did the 2311. McCollister testified
that the 2314 was "[v]ery definitely" an advance over prior disk
drives because, for example, "it had a capacity in a pack of approxi-
mately 28 million bytes as compared with 7 1/4 million bytes in an
y ;l
\
tions that had historically utilized magnetic tape or punched
card storage and in the development of new communications
oriented — or 'teleprocessing' — applications." (Tr. 90506.)
However, the demand for the 2314 disk drive "" turned out to be very
J' surprising in the rate that customers found use for it". (Case, Tr.
72743.) IBM "totally underestimated the demand for such devices" and
"we [in IBM] found ourselves hard pressed to deliver the devices as
fast as customers were demanding them" . ( Id. ) It is important to
note that the use of disk drives was not common on second generation
computing systems. According to Case, fewer than twenty percent of
computer systems prior to 1964 used direct access storage devices.
(Tr. 73527.) Nevertheless, IBM "gambled" that System/360 would be
widely used in "operational- type" applications (as opposed to batch-
type applications) and that disks would play a "pivotal role" in such
applications. (Evans, Tr. 101139.) System/360 's more advanced
operating systems, were designed in a way that required a direct
access storage device for their successful operation. The higher
jj performance and greater function necessary to achieve such operation
It
:| could not have been provided with magnetic tapes and the use of drums
j would simply have made the cost of storage too expensive. (Case, Tr.
i
j 73451-53.) IBM was therefore betting that users would be willing to
!
I
j trade-off the expense of disk drives for the increased efficiency of
i
!
j operation and the additional function that a disk-based system would
i
;
be able to provide* — that users would accept widely an approach to
* The "significance" of the disk drive was that it provided a
[functional capability of having information on-line and readily
javailable. (Rooney, Tr. 12142.) The random access capability of
t
i
I -325-
computing that had not been widely accepted before.
In hindsight, that bet was a good one. As Case testified,
today "nobody thinks of developing a wide range of computing equipment
or a family of computer systems without having a direct access storage
device as a prerequisite for the operating systems". (Case, Tr.
73452-53.) Back in 1964, however, nobody but IBM had that thought or
acted upon it as forcefully.* As a consequence, the tremendous
acceptance of IBM's disk drives swept before it all of the other
approaches to random access storage then being offered:
"During that period the entire industry and the users began
to appreciate the importance that disk drives were going to play
in the great majority of general purpose computer systems.
Before that time, alternatives were being experimented with, such
as particularly magnetic card devices, and also I think no one
realized the degree to which the transaction processing mode of
use was going to prove popular. I believe only IBM among the
major competitors at the time offered an alternative between
magnetic card devices and disk drives, with developments pro-
ceeding along both lines. A number of the other manufacturers
committed themselves almost entirely to the magnetic card devices,
sometimes also using magnetic drums.
L
2
Z
4*
5
7 !
3
?
lq
as
iz
13
i t
IS
17
IS
ta
23 ;■ disks "permitted a new and more effective approach to doing customers'
;jwork", particularly in real-time applications such as those performed
21 -| by banks and airlines. (McCollister, Tr. 9591.) System/360's empha-
ij sis on disk drives made possible more efficient use of CPU, main
22 : ; memory and peripherals; increased the range of functions and services
jthat could be provided by the operating system; and made possible a
23 {"more valuable" mode of operation (random processing of transactions)
[than the sequential access mode of processing that was common prior
24, |to the emphasis on disk drives. (Case, Tr. 73463-70.)
-c j * As we discuss below NCR, Burroughs, Sperry Rand, Honeywell and
jRCA all offered different approaches to random access storage, and
ijall of those approaches failed in the face of the tremendous user
(acceptance of disk drives. (See below, pp. 94 383, 473-74, 549-50
; !659.)
I
;i -326-
"When it became apparent that the class of magnetic card
devices was not going to be successful in the marketplace, for
reasons of reliability, and that the disk drive was a critical
product, many of IBM's competitors were left for a while without
a satisfactory option." (Withington, Tr. 56240-41.)
n
_ i
3
3
a
Both the level of performance and the attractiveness of
System/360 were substantially dependent on the 2311 and 2314 disk
drives. (McCollister , Tr. 9370, 9591-92; Rooney Tr. 12122; Knaplund,
Tr. 90506-07; Evans, Tr. 101138.) The 2.311 was "far more" important
ito the marketing of System/360 than the 1311 had been for IBM's earlier
systems, because the 2311 "offered an improved price/performance . . .
was supported to a greater degree by systems programs . . . and,
therefore, was easier to use, and . . . was more reliable". (Wi thing-
ton, Tr. 56246-49.) And the 2314 was, if anything, even more impor-
tant. It provided "a functional capability very much needed in terms
of price/performance in the competitive marketplace and without that
capability you were in a weak competitive situation against IBM" .
„ ;[ (Rooney, Tr. 12193.) Within IBM the 2314 was recognized as a "catalyst
4-
5
6
to make many systems sales for previously undeveloped application use
of computers" and as a "door opener that beats competition". (PX 1967,
pp. 1, 3, see Page, Tr. 33122.) According to Case, IBM's emphasis on the:
use of disk drives with System/3 60 contributed to the objective of
growing the market for IBM products in particular and computer system
ij products in general. (Tr. 73468-70.)
Q :!
'1 Not surprisingly, other systems suppliers wanted the kind
rr 'J
■|of "catalyst" for systems sales that IBM already had. Eventually, they
[I |
j either acquired them from OEMs or from IBM itself or they undertook to
2. :i
; manufacture them themselves. As we discuss below, the acceptance of
13 ■
•360 spurred the growth of peripheral equipment manufacturers, some of
jwhom supplied IBM 2311 and 2314 type disk drives directly to IBM end
— .1
I
! -327-
Lj users. During the latter part of the 1960s, however, these manufac-
2 : turers served as a prime source of disk drives for many systems
3; suppliers. (See pp. 753-59, below.)
4- i- Memorex was the first of the PCMs to offer IBM plug-compatible
3 if disk drives, in 1968. (See p. 770, below.) During the years
1967-70, Memorex hired almost 600 former IBM employees, three of whom
became Memorex Vice-Presidents. (JX 34, pp. 1-2.) In 1967, Memorex hired,
a number of disk drive engineers from IBM, including Roy Applequist,
who had designed IBM's voice coil actuator. (Guzy, Tr. 32858-64;
Gardner, Tr. 38585, 39143*) Applequist designed the voice coil
actuator for Memorex 's 630 disk drive, which, according to an indepen-
dent engineering assessment, was "directly derived" from IBM's 2314B
(3330) and "not the result of coincidence". (Gardner, Tr. 39143; DX
1418, p. 151; see also Spitters, Tr. 55259-61; DX 2572.) D. J. Guzy,
former Executive Vice President of Memorex, testified that the hiring
^g -{of Applequist and other IBM engineers was important to the success
£7 ; ithat Memorex achieved with the 630; and that the 630 and 660* were
'I
IS , styled and intended to be, respectively, 2311-type and 2314-type disk
19 j drives. (Tr. 32316, 32776, 32899.) Memorex marketed the 630 and 660
2G i not only directly to IBM end users, but also to a number of different j
■' i
21 J systems manufacturers, including RCA, Univac, DEC, Burroughs, Honeywell,;
■i i
22">SEL, Hewlett-Packard, Siemens, Phillins and ICL. (Guzy, Tr. 33168; j
23 jDX 1302, pp. 1-3; DX 1308, p. 1.)
24 1
:i
2z I * Memorex did not begin volume production of the 6 60 until the
i second quarter of 1969. (DX 1268, 0. 17.)
:i
•! -328-
7
3
9-
10
II !
12!
13 •
14
1 I
2 I
3 !
4
5
6
7
8
9
10
11
j
12 |
i
13 i
15
16
ISS was formed in December 1967 by twelve former IBM employ-
ees who had resigned from the San Jose Laboratory, where they were
responsible for disk drive development. A number of this so-called
"dirty dozen" had worked on IBM's Merlin (3330) program. (Whitcomb,
Tr. 34555-56; DX 4756B, p. 96; DX 4739: Wilmer, Tr. (Telex) 4266; DX
4741: Yang, Tr. (Telex) 6116.) Like Memorex, ISS manufactured 2311-
type and 2314-type disk drives, the 701 and 714, which were marketed
by Telex to IBM end users beginning in 1969. (PX 4732A, p. 12; DX
4242, p. 8; DX 4250, p. 7; DX 4756A, pp. 36, 7 2.) ISS also marketed
disks OEM to Hewlett-Packard, Itel and Storage Technology Corporation.
(DX 86A, p. 2; DX 4113: Terry, Tr* (Telex) 3310-12.) The ISS 2311-
type drive was similar to IBM's 2311 except for the addition of a
voice coil actuator, and the ISS 2314-type drive was functionally
14 j equivalent to IBM's 2314, again except for the addition of a voice
coil actuator. (Page, Tr. 33072-73; Ashbridge, Tr. 34812-13.) ISS
was eventually acquired by Sperry Rand (in 1973) for its advanced
17 ;|disk technology, technical capabilities, highly qualified personnel,
18 j
19 |
20 j
i
21 j
plant facilities and highly profitable OEM customer base. (DX 86A,
pp. 1, 4, 5; DX 87, p. 12.) After the acquisition, ISS became the
developer and manufacturer of disk subsystems for use in Univac
systems, but continued marketing 2314-type disk drives to IBM users
22 ! iand to OEM customers. (Eckert, Tr. 988-89; McDonald, Tr. 4060-63.)
23 i
!■
24 I
i
25 i
CalComp also offered 2311-type and 2314-type disk drives,
manufactured by Century Data Systems, to end users and on an OEM
basis. (DX 10735, pp. 10-11; see pp. 776-777, below.) CalComp
shipped its first plug-compatible (2311-type) disk drive in June 1969
-329-
l'; (PX 5324, p. 46; DX 4756A, p. 8), and later became the "first company
2'i to produce and ship a 2314 equivalent". (PX 3707A, p. 38; DX 10735,
2 T p. 10.) Century Data marketed these disk drives to leasing companies
i
^ |. such as Randolph and to other systems suppliers such as Nixdorf,
!h Burroughs and Univac (PX 3146A, p. 1; PX 5581, p. 10; PX 5582, p. 7;
![ DX 1886, p. 7; DX 12194.)
Similarly CDC manufactured and marketed 2311- and 2314-type
disk drives, both end-user and OEM. CDC's OEM customers included
Honeywell, GE, Siemens, RCA, XDS, ICL, SAAB, CI I, Burroughs and
Telex. (G. Brown, Tr. 51056-57, 51080-81, 51095-96; see pp. 682-84,
1074-77, below.)
RCA did not even wait for PCM's to copy IBM's technology, but
went directly to the source. "It was apparent [to RCA] that this
capability which was offered by IBM was going to be required by RCA
in order to successfully market its products."
/
3
Si
10
u
12
13
14-
"This capability at the time was not available from any other
IS ;1 source. So, therefore, when we announced the Spectra 70 family
it or series, which came out about eight months after the IBM 360
17 i announcement, we announced as a part of the RCA product line this
' particular Model 2311 disk pack file capability and we obtained
15 ! these files by buying them from IBM, the same as any other
;j customer would buy them from IBM." (McCollister , Tr. 9370.)
19 :
Although RCA had its own disk drive development program, RCA
subsequently contracted with Memorex to supply disk drives for use with |
!
RCA computer systems because Memorex' development program was further
ahead than RCA's "which was going to be about a year later than !
Memorex' s". (Beard, Tr. 8575.) RCA went to Memorex at a time "when !
i
we had in parallel our own development going on" because RCA was "under •
-330-
parable product to the IBM 2314 at the time". RCA "couldn't afford
in the marketplace to wait that additional year" necessary for RCA's
development program to produce the required disk drives " [bjecause
we were losing too many sales for the lack of it" to IBM. ( Id. )
GE, on the other hand, attempted to build an IBM plug-
compatible 2 3 11- type drive. (Ashbridge, Tr. 34 812-13; G. Brown, Tr.
51536-37; Spain, Tr. 90227.) But "it met with limited success and
arrived to the marketplace much too late to meet market, or customer
requirements". (G. Brown, Tr. 51536.) GE entered into an exclusive
contract with Greyhound Computer Corporation to sell the device, but
Greyhound ended up having to take a significant write-off on its
investment in the GE equipment and even sued GE. (Spain, Tr. 88753,
88755.)
Not until the very end of the 1960s had IBM's disk tech-
nology been sufficiently spread around the industry for some of
IBM's systems competitors to have pulled even. Thus, the January 5,
1970 Phase III Level Forecast Assumptions for IBM's soon-to-be
announced Merlin* disk drive reported:
'j "System Manufacturers
j [
;| "From the announcement of the 2314 in 196 5 until late in 196 8 I
'! IBM had significant competitive advantages in this product area, j
■j as no competitor could offer a direct access device with the j
\ price, capacity, performance, and interchangeability character- j
j istics of the IBM 2314. The situation today, however, has changed j
; ; radically as most system manufacturers now have announced devices j
j which are virtually identical in specifications to the IBM 2314. j
; : The chart below tabulates the status of the ten major system j
* As we shall see, the Merlin (3330) drive put I3M right back in
the lead in disks. (See below, pp. 898-902 . )
-331-
L
2 *
4.
Si!
8 ;
10!
i
i
II ;
i
T -2 : ;
14- ;
I
lS i
1
IS ;
17 I
IS :
19
■■\
20 'i
1
21 j
-i
^2. ;!
1
23 :|
24.
25
manufacturers in this regard.
Marketed By
Mf g . By
2314 Type Media
Status
Burroughs
Burroughs
No
Fixed Disc
Delivered
CDC
CDC
Yes
2316
Announced FCS 2Q70
DEC
Memorex
Yes
2316
Imminent Delivery
GE
IBM
Yes
2316
Announced
Honeywell
Honeywell
Yes
2316
Announced FCS 2Q70
IBM
IBM
Yes
2316
Delivered FCS 1Q67
NCR
NCR
No
Strip
Delivered
RCA
RCA
Yes
2316
Announced FCS 1Q70
SDS
Memo rex/CDC
Yes
2316
Imminent Delivery
Univac
Univac
Yes
2316
Announced FCS 1Q70
"The rental prices offered by CDC, GE, Honeywell, RCA, and Univac
are within a few percentage points of the IBM 2314. (CDC and
Honeywell discount by approximately 10% for three to five-year
leases.) Burroughs and NCR use radically different approaches
and price comparisons cannot be weighed properly due to the
lower performance levels of their devices. To date, competitive
system manufacturers have not had any significant price advan-
tage in the file facility environment." (DX 7858, p. 2.)
(viii) Standard Interface /Modularity. IBM adopted a
"standard interface" for the peripherals in the compatible 360 line.
This meant that (with some exceptions*) the same peripherals would
* Such exceptions as existed came about as a result of design
trade-offs. Some peripherals such as the 2301 and 2303 drums with
high speeds, for example, were not made attachable to the slower
models of System/3 60 (such as the Models 20, 22, 25 and 30) because
those smaller CPUs could not accept the high data rates of these
peripherals. (Case, Tr. 73449-50.)
In some instances (such as with the 360/25) peripherals were
attached directly to the CPU rather than through the standard inter-
face because designing a "native attachment, closely integrated with
the computer", provided "somewhat greater performance at somewhat
lesser cost". (Hughes, Tr. 71941; Case, t Tr. 73450; see also PX
2209A, pp. 15, 17.) In such cases, of course, the cost/performance
improvements were achieved at the expense of some of the configuration
flexibility that was afforded by the standard interface. (Hughes,
Tr. 71941-42, 71995.) The dilemma of when to make such trade-offs
was a difficult one both during the development stages of System/36
(see Gardner, Tr. 38387-88, 38958-61, 39110-13; DX 1656, DX 1657,
DX 1658, DX 1659) and thereafter. (See Haughton, Tr. 95019-24;
DX 1662.)
i r
-332-
1
2
3
4
5
6
7
S
*
LQ
11
12!
13
14
15
16 .
17 !
13 =
attach to ail processors in the line and would do so in the same
way. The standard interface, together with compatibility, helped
maximize the benefits that customers could derive from the broad
range of peripherals offered with 360 and the compatibility across
the entire line. It helped give System/360 a configurability that
was unmatched by competitors and permitted customers the utmost
flexibility to optimize their data processing systems by piecemeal or
modular changes. At the same time, it enabled IBM to reduce costs
through economies in development and manufacturing. Others
undoubtedly recognized these benefits and also moved toward more
modular product lines — but not until well after IBM had done so.
(Case, Tr. 73446, 73474-75, 73523.)
The requirement for a standard interface for the New Product
Line was implied by two of the architectural and engineering "ground
rules" set out in the SPREAD Report — i.e., that "all channels shall
appear identical" to any I/O device type and that -"the I/O gear shall
not need to be changed" when one processor is substituted for a
slower one. (DX 1404A, pp. 19-20 (App. A to JX 38).) Case and
iq :| Hughes testified that the standard interface became a "design objec-
2Q :j tive" for System/360. (Hughes, Tr. 34102-03; Case, Tr. 73446.) Case
1
3«j I explained:
22 ■■ "[W]e had as an objective to design a number of different peri-
•i pheral devices that would each be able to plug into central
2- ! processing units of the whole System/36 family. We wanted to
| do this in a way which would maximize the degree of choice that
* A \ customers would have in selecting peripheral devices to go with
""* j central processing unit models, and to do it in a way which
m S j would minimize IBM's development expenses in designing those
-333-
L
2
3
4-!
5
7
3
ia
Li
12
14
peripheral devices, and do it in a way which would help us to
reduce our manufacturing costs of the peripheral devices by
achieving as large as possible a production run of each par-
ticular device.
"The technique that we chose to accomplish these objectives
was called the System/360 channel to control unit interface,
often abbreviated with the words 'standard interface 1 ". (Case,
Tr. 73446.)*
The standard interface, together with compatibility, provided
IBM with a number of development and manufacturing advantages. "It
reduced the design time of many groups" who would otherwise have
spent time designing their "own pet means of attachments". (Hughes,
Tr. 71939.) Instead, the CPU and peripherals designers were able to
concentrate on building "the best products they knew how" and on
"advancing the state of their art as far as possible". (Case, Tr.
73447.)
The standard interface, together with compatibility, also
helped IBM reduce ' development costs by reducing the number of circuits
that had to be designed to permit each peripheral to attach to each
j CPU. Prior to System/360, peripherals that attached to the central
j processing unit did so by means of a unique interface. As a result,
[ a separate design effort and set of circuitry was required for each
such attachment to the central processing unit. With much of System/
360, only a single design effort and set of circuits was required
17
is i
r
ZQ 'i
.1
21 ■! • !
j * The control unit to peripheral device interface was not stan- •;
23 ! dardized, however, which meant that each device required its own con- •
j trol unit. The objective of the New Attachment Strategy in the 1970s
2i, ; was to standardize the device to control unit level interface and j
I thereby achieve benefits similar to what had been obtained with the ;
?q I standardization of the control unit to channel interface in System/360. !
! (Case, Tr. 74079-83; Haughton, Tr . 95010-32.) " j
! i
i '
1 '
'■\ -334- !
because of the standardization of the interface between the control
unit and the channel of the central processing unit. (Case, Tr.
73446-48.)
The standard interface, together with compatibility, helped
simplify and cost reduce IBM's manufacturing process. "[I]t led to
higher quantity production runs of the peripheral devices since the
same peripheral device and the same attachment, or plug-in circuitry,
was associated with the interface to any of the CPU models". (Hughes,
Tr. 71939-40; Case, Tr. 73448.) Because of this commonality, similar
economies were achieved in the testing process. That was particularly
important to IBM in getting 360 ready for announcement. Hughes
testified that
"since we had a multitude of I/O devices and a prescribed time
to get it done, [compatibility and the standard interface] helped
us a great deal in both our engineering and all aspects of
testing ... to get the total job done". (Tr. 71939-40; see
also Case, Tr. 73533.)
Case testified that a related objective of the 360 Advanced
I Systems Group was to develop "elements of a computer system which
| could be put together, or configured in a wide variety of ways".
I
| (Tr. 73416.) That objective, which Case called "modularity", was
i
:j promoted by the standard interface because it allowed users to plug
1 any peripheral device into different 3 60 central processing units
!
:j "without changes in the central processing unit". (Case, Tr. 73448;
i
j see also Hughes, Tr. 34109.)
i Not only did IBM achieve the modularity objective set for
.j
| System/36 (Case, Tr. 73420) , it did so to an extent that other manu-
I facturers were unable to match for almost a decade. Among the manu-
i
I -335-
L ;
2
3
4-
5
a
7
3
9
10
II
IZ
13
14.
« ^»
15
?T
facturers and marketers of computer systems from 1964 to 1972,
"IBM was the leader in providing . . . modularity. [*] With the
announcement of the System/360, IBM provided the first line
offering anything like the degree of modularity which has since
become available from all the major manufacturers.
"During the 1960's, all of the manufacturers, including IBM,
evolved their product lines further in the direction of making
them more modular, but ... it is fair to say that throughout
the period . . . IBM's product line remained the most modular
of all the general purpose product lines available". (Withington,
Tr. 58268-69.)
Accordingly, System/360 's modularity provided benefits to users that
were unavailable from competitors and provided an incentive to acquire
36 0s that did not exist with respect to competitive systems. As Case
testified:
"The achievement of the modularity objective was . . . very
helpful to IBM in enabling the computer products produced by
IBM to be chosen by customers in a way that would optimize
the price/performance of their installation, and in a way which
would provide for convenience and small accepted changes in
the installation as the requirements of the enterprise changed.
"That is an important benefit to customers for two reasons:
"First, . . . they can most accurately adjust the capabili-
] ties of their computing installation and, hence, the cost to ' .
17 ; them of their computing installation to their real needs.
XS I "Second, . . . they are able to change the performance or the
;j capabilities of their configuration to match their changing
£g i requirements . . . without changing the entire installation, but
;j just adding or subtracting parts, or boxes from the installation."
?0 "• (Tr. 73427-28; see also Navas, Tr. 41394-95; Withington, Tr.
! 56193.)
* " [A] modular line of computer systems is one in which every element;
22 \ of the system, including processor, storage, peripheral equipment, and j
systems programs can be independently exchanged for a compatible larger I
or successor module in such a manner that over time the installed com- j
puter system may evolve to a much different or a much larger or a much :
more capable one without any particular point in time being identifiable!
as one in which the entire system was converted from one to another." i
(Withington, Tr. 58268, see also Tr. 58269-76.) I
-336-
IBM's achievement of modularity for System/360 "helped
to remove limitations on the use of computing equipment that had pre-
viously existed" because it relieved users of the need to make "system"
. changes. (Case, Tr. 73435-37. ) IBM, more than any other firm,*
reaped the benefits of user demand for modular acquisition alterna-
tives :
"Because the achievement of the modularity objective was useful
for customers, it was of benefit to IBM in that.it tended to
increase the value of IBM products as compared to the products
of others, and with an increased value, our sales tended to
increase and that was important in the achievement of the total
success, or the total order rate for System/360 computers and
the peripheral devices that were part of those computing systems."
(Case, Tr. 73428.)
There were, however, risks associated with modularity and
the standard interface. The design trade-offs necessary to create a
system which could be assembled in a wide range of configurations,
might have resulted in a design that was not optimal for any particular
configuration, at a cost higher than it need otherwise have been.
\ j Development of the standard interface entailed a similar risk "that
no one attachment or no one plug-in capability [would be] optimal for
: ; * Other companies followed IBM's lead in making their product lines
■| more modular, but were not as advanced or fast moving. For examnle,
<, '!
■i '
! a) Modularity "was beginning to appear" in Honeywell's line
I ! by approximately 1966, but it "was still far less than
j available in the IBM line" and did not "span the range
1 \ of available modular options that IBM's line did" through
" j the 196 0s;
| b) By 1977, Univac ' s line was "probably still deficient"
i j compared to IBM; and
s "\ c) "Burroughs' modularity was restricted by the narrowness of
! its product line . . . through most of the 1960s".
| (Withington, Tr. 53271-75.)
!
! -337-
L| the particular device involved". (Case, Tr. 73531-32.) Thus, the
2 : question of separate control units versus native attachment of peri-
i
2 ; pherals became a matter of some controversy within IBM, involving
4*! important dissenters (such as Haanstra) from the stand-alone control
J !• unit method of attachment which was finally adopted for most of 360.
a!! (See DX 1656; DX 1657; DX 1658; DX 1659.)
7 ij There was risk to IBM of another type as well. 360' s
standard interface and modularity of design, together with its wide-
ranging compatibility, presented an attractive target for competitors.
The new, modular environment in which 360 would be offered created
the prospect that other manufacturers would produce "modules" that
would be marketed in direct competition with comparable IBM products .
The standard interface of System/360 offered others the same advan-
tages it gave IBM* — and more. As Case testified,
"It reduces their design costs as it did for IBM, and it allows
them to achieve higher production runs as it did for IBM, and
it allows users to conveniently plug in peripheral devices of
their manufacture just as it allows the convenient plug-in
of devices of IBM manufacture". (Tr. 73474-75; see also Navas ,
Tr. 41395-96.)
-0 I * That was particularly true because IBM published a number of manu-
| als which were readily available "at a nominal charge of a couple of
*- '{ dollars" and which described the mechanical, electrical and logical
•j characteristics of IBM's interfaces in a way that permitted manufac-
22 .j turers of peripheral devices to design "workable and safe" attachments
! of their devices to an IBM channel and which permitted CPU manuf ac-
-^ j turers to attach IBM peripherals to their own CPUs in a like manner.
'(Shoemaker, Tr. 30867; Case, Tr. 74125-50; Peterman, Tr. 99441-43; DX
2- 17590, Perkins, pp. 21, 24; DX 7591, Hilyer, p. 15.) IBM's OEMI
| (Original Equipment Manufacturers Information) Manual for System/360
25 jwas first made available in 1965. (Case, Tr. 74145.)
-338-
r
i
of being able to copy IBM's designs and use IBM's software without !
having to invest in developing either. As a consequence they could
be expected to have lower costs than IBM and to offer their products
at lower prices than IBM initially charged. (Case, Tr. 73523; Cary,
Tr. 101333-37, 101339, 101374, 101629-31; see also Wright, Tr.
J 13236-38; Enfield, Tr. 20765-68; G. Brown, Tr. 51812; Powers, Tr.
95376-89, 95412-13, 95475-82; PX 3312A, p. R14; PX 3594A, pp. 4, 26, 36,
40; PX 3681A, p. R-l; PX 4880, p. 3.)
The prospect that others would be able to "tap" IBM's
support and offer compatible products in competition with IBM was
foreseen by the SPREAD Committee and others within IBM prior to 360 's
announcement. (Knaplund, Tr. 90497-98; DX 1404A, p. 40 (App. A to
JX 38); see also PX 3908A.) That prospect became a reality in the
late 1960s and in the 1970s — with numbers of competitors offering
replacements for each and every box in IBM's systems. IBM could not
keep to itself the advantages of compatibility, modularity and the
standard interface.* On the other hand, IBM really had little
:j * John Navas of Memorex explained the benefits for a manufacturer
' :j of plug-compatible products in being able to attach a single disk
,| drive model to a variety of 360 processors:
T !
t
"From the standpoint of a company such as Memorex, it would
tend to reduce product cost to minimize the number of models of
•I a given type of disk drive which we would be producing. That
- '.| would result in a higher production volume for each type of
unit, and would result in less development expense associated
with developing the various models ....
"If Memorex had had to produce unique models of its 630 for
each of the various models of IBM System/360 ... it would have
probably increased the development expense, caused an increase
in manufacturing costs, and increased the difficulty and adminis-
trative expense associated with lease base management". (Tr.
41395-96.)
-339-
L
Z
z
4-
f
a
alternative but to provide such features if 360 was to succeed. It
was a matter of responding to "a competitive necessity". Because of
user demand, "the manufacturers attempting to compete were forced to
maintain continuous developments of different modular types of equip-
ment that could be configured together". (Withington, Tr. 56174.)
However, the great modularity of System/360 meant that IBM
7 j would have to price each and every box in the system carefully.*
8 : According to Knaplund, IBM had to make those prices attractive on a
9 j box-by-box basis because users made box-by-box performance comparisons
i
IQ ; between IBM and its competitors; because System/360 was susceptible to
1^ | such a wide range of configurations that a single box price that was
12.1 out of line could make the whole system unattractive; and because
< .
12. '■■ competition was anticipated from suppliers of plug-compatible peri-
!£ !( pherals and CPU's who would attempt to replace IBM's products on a
15 ! box-by-box basis. (Knaplund, Tr. 90496-98.) That last reason, in
^g 1 particular, made competitive box prices for System/360 "critical".
« T i (Id.)
19 1
21 f
j * IBM has always priced its products on a box basis, with each unit
7- ; of EDP hardware (such as a CPU, tape drive, disk drive or terminal)
"" j offered at a consistent price regardless of the type or number of
~s 1 boxes that a user combined to configure his system. (Knaplund, Tr.
*" :j 90495-500; Akers , Tr. 96665, 96675-76; Cary, Tr. 101386-37.)
-340-
21 \ . (
r
Appendix
Examples of System/360 Uses
The following are some of the diverse applications for
which System/3 60s have been used:
By a French research and consulting firm to study ways of
increasing the power output of large hydroelectric dams (DX 13677,
p. 16) ;
By a petroleum exploration company to prepare seismic
reports ( id. , p. 14) ;
By a manufacturer of animal feed concentrates for feed
formulization (DX 13678, p. 9);
By the Deutsches Elektronen Synchrontron in Hamburg,
j Germany, to evaluate photographs of bubble traces left by invisible
elementary particles in an electron accelerator (DX 13679, p. 20);
By a Japanese steelmaker for automatic control of the
steel manufacturing process ( id. ) ;
By Swissair for automated message switching and automatic
| passenger check-in and weight-and-balance calculating ( id. ) ;
*
\ By a paint manufacturer to signal corrections for
* i
:j deviations in ingredients and production cycle ( id. , p. 10) ;
.1 By scientists in New England to simulate and study the life
I '•
i cycle of lobsters ( id. ) ;
m
j By African Ivory Coast harbor authorities to compile and
! analyze statistics on tropical wood exports ( id. , p. 13) ;
A \
j By a Swiss chemical manufacturer to operate an automated
-Al-
3
s
7
3
Q
10
u ;
iz I
!
I
n J
L4*
15
15
warehouse (DX 13680, p. 28);
By BOAC to calculate tariffs, management statistics and
flight plans ( id. ) ;
4^i[ By Japan's national broadcasting company to maintain
schedules and budgets for 640 television shows and 1,200 radio
programs, and to control actual broadcasts ( id. , p. 16) ;
By IBM's Field Engineering Division for computer assisted
instruction (DX 3364, p. 9);
By an air freight company for instantaneous tracking of
daily shipments ( id. , p. 22) ;
By the architectural department of a county council in
England to design municipal buildings ( id. , p. 24) ;
By a supermarket chain to calculate unit prices (PX 5767,
il P- 13);
By American Airlines (3 60/65) for airline reservations
(Welch, Tr. 75385-86) , field maintenance reliability applications
•| (O'Neill, Tr. 75848-53) , crew qualification and takeoff power assist
17
(id., Tr. 75909-10) , flight planning (id. , Tr. 75928) , and calcula-
; tion of estimated time of arrival ( id. , Tr. 75976) ;
i3 i
•; By Aspen Computype, Inc. (3 60/4 0) for typesetting
2C ■!
'! (DX 6078, McCaffery, p. 9) ;
2i "i .
j By Autocomp, Inc., (360/40 and 360/50) for typesetting
\ (DX 4039, Kendall, pp. 7-3);
23 i ,
! By AVCO Computer Services m Wilmington, Massachusetts,
~A \
*"" I (360/75) for:
-A2-
1 ! i
2 !|
3 j!
5 !
1
6 :f
7 I
ii
8 j
9 l
10 j
ii i
12 |
13 1
.!
.1
14 ||
15 ii
16 =i
17 !:
13 : ;
19 :|
20 '!
21 ■
22 •
23 i;
24. ;
25 ;
drafting applications
FORTRAN flowcharting
geometric design
mathematical functions
frequency distributions
movie making
perspective plotting
trajectory analysis
financial analysis
production control
statistical analysis
mathematical analysis
applied statistics
structural load analysis
structural shell analysis
structural ring and
frame analysis
antenna pattern
prediction
communication link
analysis
plasma attenuation
analysis
drag coefficient
analysis
aerodynamic heating
analysis
heat transfer
analysis
thermochemical
equilibrium analysis
flow field analysis
boundary layer
analysis
penetration aids
analysis
decoy model analysis
radar cross section
analysis
finance applications
manufacturing
applications
(DX 6816, pp. 3, 10, 12, 13) ;
By Bowne Timesharing, Inc. , (36 0/40 and 3 60/50) for time-
sharing text editing (DX 6090, Abrams , pp. 9-10);
By Carnation Corporation (360/40) for telecommunication
-A3-
L
2
3
4-
3
5'
7
S
ia
Ll
12
13
14
is:
17 i:
is
applications and linear programming (Navas, Tr. 39177-78);
By Computone Systems, Inc., (360/50) for architectural
design and mathematical modeling (DX 4069, Robeson, pp. 16-17);
By Continental Illinois National Bank & Trust (360/50) for
on-line credit authorization (DX 4756, p. 7);
By DP&W, Inc., (360/30) for business and engineering
applications (DX 4076, DiPietro, pp. 8-9);
By the San Francisco Federal Reserve Bank (360/50) for
message switching (Withington, Tr. 57540; DX 2667, p. 3);
By the Fluor Corporation (360/50) for project planning and
control, process simulation, process analysis, refinery simulation,
structural design, piping design, electrical design and mechanical
design (DX 4023, Neher, pp. 11-12, 17);
By General Motors Research (360/50) for timesharing (Hart,
Tr. 80505-08);
By the New York Police Department (360/40) for automated
dispatch and identification of police vehicles (DX 4756A, p. 58);
By the Orange Coast College District in Costa Mesa,
fcalifornia (360/50) for computer assisted instruction, grade report-
1S ;
ling and student registration (King, Tr. 14 761-62) ;
;| By Pacific Southwest Airlines (360/65) for passenger service
""applications (O'Neill, Tr. 76019);
22 ''•
i 3y Proprietary Computer Systems i Inc., (360/65) for:
:j
25 l -
!
I
I
1
■! -A4-
banking services
accounting
manufacturing control
three dimensional COGO
stress analysis
digital signal processing
reliability calculations
electrical engineering
fast fourier transforms
matrix analysis
chemical engineering
graph plotting
thermal analysis
linear programming
CPM analysis
PERT analysis
Monte Carlo analysis
Markov analysis
integration
differentiation
non-linear equations
regression analysis
descriptive statistics
transducer calibration
(DX 3960, Barancik, pp. 11-12);
By Pyramid Industries, Inc., (360/40) for time shc.ring
(DX 4756D, p. 23) ;
By Southern Railway (360/50 and 360/65) for on-line
monitoring of railroad cars (DX 4756D, p. 42) , (360/50) for peripheral
processing (J. Jones, Tr. 79848, 79413-14), (360/30) for card to tape,
tape to card, and tape to print processing, and peripheral processing
( id. , Tr. 79243);
By TBS Computer Centers Corporation (3 60/30 and 360/40)
for data communication, remote teleprocessing, accounting reports and
statistics, inventory, cost analysis, market research, production
control, accounts receivable and payable, traffic studies and order
-A5-
L
2
2
4-
a
a
7
3
9 !
ia
U !
12. 1
analysis (DX 7134) ;
By Union Carbide (3 60/30) for message switching (McGrew,
Tr. 77271) .
System/360 's uses within the Federal government alone
illustrate graphically the broad range of applications performed by
360 users. For example, 360s have been used:
By the Headquarters, U*S. Marine Corps (360/20) , for
"Automated Communications Processing System" (DX 2992,* pp. 619,
1123-1125) ;
By the Veterans Administration, Austin, Texas (3 60/20) ,
for "Patient Care" (DX 2992, pp. 1073, 1158);
By the Veterans Administration, Washington, D.C. (3 60/20) ,
for "Facility Planning and Construction" and "Fiscal Accounting"
: (DX 2992, pp. 1078, 1158) ;
By the Veterans Administration, Philadelphia (360/20) , for
"Insurance" (DX 2992, pp. 1076, 1158);
IS ; i
By the Defense Nuclear Agency, Headquarters, Field Command
14-
17
15
19
(360/20) , for "Logistics - Supply" and "Stockpile Management" (DX
2992, pp. 546, 1121; DX 4593, p. 133) ;
\ By the Department of Air Force, Air Force Systems Command,
axil
Los Angeles, California (360/20), for "Telecommunications" and
i
fT .1 '
— -• !
: ] "Command and Control" (DX 2992, pp. 452, 1120; DX 4593, pp. 103, 104)
'i
23 \
J * DX 2992 is the Stipulation and Amended Response of Plaintiff to
24. 'J IBM's Interrogatory 5(e). Examples of applications taken from DX
j 2992 are described here in the same terms in which they are described
25 j in DX 2992.
■i - -A6-
f i
i
» '
i !
By the Atomic Energy Commission, Division of Technical
Information (360/20) for "Operations Control and Support" (DX 2992,
pp. 118, 1113; DX 4593, p. 72);
By the Atomic Energy Commission, Oak Ridge Office (360/20) ,
for "Scientific and Engineering" (DX 2992, pp. 91, 1113; DX 4593,
| p. 61);
By the- Department of Commerce, Office of the Secretary
(360/20), for "General Administration" (DX 2992, pp. 149, 1117;
DX 4593, p. 74) ;
By the Department of Commerce, Bureau of the Census (360/20) ,
for "Statistical Programs" (DX 2992, pp. 157, 1117; DX 4593, p. 77);
By the Department of Air Force, Air Force Systems Command,
Eglin AFB (360/20), for "Research, Engineering" (DX 2992, pp. 442,
1120; DX 4593, p. 101) ;
By the Department of Air Force, Air Force Communications
Service, Offutt AFB (360/20), for "Weather, Environment"" (DX 2992,
pp. 411, 1120; DX 4593, p. 94);
By the Marine Corps Headquarters, FMFLANT (360/20), for
"Automated Communications Processing System" (DX 2992, pp. 6 31,
1123-25) ;
By the Department of Navy, Naval Intelligence Command
(360/20), for "Intelligence Data Handling System" (DX 2992, pp. 733, !
I
1123-25; DX 4593, p. 156); |
I
By the Department of Navy, Naval Research Laboratory j
(360/20), for "Laboratory Support Systems" (DX 2992, pp. 610, 1123-25) ;j
-A7-
L
2!
3
±\
a
7
3
ia
u
12
12
14
15
IS
17
IS
By the Department of Navy, Commander Naval Reserve
(360/20) , for "Navy Manpower and Personnel Management Information
System" (DX 2992, pp. 657, 1123-25; DX 4593, p. 138);
By the Department of Navy, Pacific Fleet Commander in
Chief (360/20), for "Air Logistics Support Systems" (DX 2992, pp.
722, 1123-25; DX 4593, p. 152);
By the Defense Supply Agency, Lemoncove, California
(360/20), for "Communications" (DX 2992, pp. 802, 1126);
By the Export/Import Bank of the U.S. (360/20), for
"Payroll and Personnel", "Accounting" and "General Administration"
(DX 2992, pp. 818, 1127);
By the National Aeronautics and Space Administration,
Goddard Space Flight Center, Greenbelt, Md. (360/20) , for
: "Scientific" and "Engineering" (DX 2992, pp. 907, 908, 1144);
By the National Aeronautics and Space Administration, Jet
Propulsion Laboratory, Pasadena, California (360/20) , for "Business-
i
j Commercial" (DX 2992, pp. 937, 1144);
| By the Department of Treasury, Office of Treasurer (360/20)
I for "Administration of Government Finances" (DX 2992, pp. 1066, 1155;
| DX 4593, p. 194) ;
ZQ "I
! By the U.S. Defense Communication Agency (3 60/20 and
21 '•
j 360/30), for communication control and as terminals (DX 7524,
22 '[
j Levine, pp. 34-36, 57);
23 ';
I By thts Atomic Energy Commission, Brookhaven National
i Laboratory (360/30) , for "Personnel Management" and "Operations
-A8-
Control and Support" (DX 2992, pp. 6, 1113);
3y the Civil Aeronautics Board (36 0/30) for "Traffic
Capacity", "World Benefit Study", "Air Cargo" and "Payroll, Manpower
Distribution" (DX 2992, pp. 134, 1116);
»
By the Department of Army, Air Defense Board (360/30) ,
for "Research, Engineering" (DX 2992, pp. 174, 1120);
By the Department of Air Force, Air Force Finance Center
(360/30), for "Finance, Accounting", and "Payroll, Benefits" (DX 2992,
pp. 279, 1120) ;
By the Department of Air Force, Air Force Systems Command
(360/30), for "Research, Engineering" (DX 2992, pp. 433, 1120;
DX 4593, p. 99) ;
By the Defense Communications Agency, European Area
(360/30), for "Communications Control and Management" (DX 2992, pp.
550, 1122; DX 4593, p. 133);
By the Defense Nuclear Agency, Headquarters Field Command
(360/30), for "Test Command", "Accounting and Finance", "Communica-
tions Processing" and "Data Automation" (DX 2992, pp. 54 6-47, 1121;
DX 4593, p. 133) ;
: !
' ;; By the Department of Labor, Bureau of Labor Statistics
} ,j (360/30), for "Statistical and Economic Survey Appl." (DX 2992,
:i
L ! pp. 884, 1142; DX 4593, p. 175);
:■!
- :j By the National Aeronautics and Space Administration,
5 Johnson Space Center (360/30), for "Scientific" and "Business-
Commercial" (DX 2992, pp. 971, 1144);
-A9-
L j
Z.\
z !
3 if-
a
7
3
9
10
li
12
13
!A
15
la
17
i
IS
20 !
•i
21 I
•I
22
23 i
2± j
!
By the National- Aeronautics and Space Administration,
Goddard Space Flight Center (360/30) , for "Mission Control" and "Data
Reduction" (OX 2992, pp. 907-8, 1144);
By the Tennessee Valley Authority, Computing Center Branch
(360/30), for "Power Supply and Use", "Fertilizer and Munitions
Development", "Financial Management" and "Personnel Management" (DX
2992, pp. 1068, 1156; DX 4593, p. 194);
By the Securities and Exchange Commission, Office of Data
Processing (360/30), for "Mass Information Storage and Retrieval",
"Statistical and Economic Analyses" and "Administrative Processing
(Personnel, Payroll, etc.)" (DX 2992, pp. 1025, 1151, DX 4593, p. 191);
By the Department of Transportation, Transportation Systems
Center (360/30), for "Financial Administration", "Inventory, Supply
and Logistics" and "Planning, R and D" (DX 2992; pp. 1034, 1154;
DX 4593, p. 165) ;
By the Social Security Administration (360/30) for real
time claims tracking and real time tape library control (DX 5 792,
pp. 17-18);
By the Department of Navy, Marine Corps - COMCAB West
(360/30), for "Automated Communications Processing System" (DX 2992,
pp. 571, 1123-25);
By the Department of Navy, Chief of Naval Operation (36 0/30) ,
for "Space Surveillance System" (DX 2993, pp. 669, 1123-25; DX 4593,
p. 141);
3y the Department of Navy, Ordnance Systems Command (36 0/30) ,
-A10-
for "Ordnance Support System" (DX 29 92, pp. 694, 1123-25; DX 4 593,
p. 145) ;
By the Defense Supply Agency, Assistant Director of Plans
(360/30), for "Logistics - Depot Level" and "Communications" (DX 2992,
pp. 812, 1126; DX 4593, p. 135);
By the Defense Supply Agency, Assistant Director of Plans
(360/30), for "Headquarters Management" (DX 2992, pp. 815, 1126;
DX 4593, p. 136) ;
By the Federal Deposit Insurance Corporation, Division of
Research (360/30), for "Economic Research" ,. "Fiscal Accounting",
"Bank Liquidation" and "Bank Statistics" (DX 2992, pp. 823, 1130;
DX 4593, p. 171) ;
By the Department of Health, Education, & Welfare, Food and
Drug Administration (360/30) , for "Disease Prevention and Control" and
"Consumer Protection" (DX 2992, pp. 855, 1137; DX 4593, p. 173);
By the Atomic Energy Commission, Albuquerque Office
(360/40), for "Material Management", "Facilities Management",
"Operations Control and Support" and "Scientific and Engineering"
(DX 2992, pp. 35, 1113; DX 4593, p. 46);
i
I By the Department of Commerce, National Oceanic and
n ';
* I Atmospheric Administration (360/40), for "Mapping, Charting and
a
7
j Marine Description" (DX 2992, pp. 148, 1117; DX 4593, p. 74); j
^ i
By the Department of Commerce, National Oceanic and
i- i
a j Atmosphere Administration (360/40), for "Environmental Prediction
j
•^ j and Warning" (DX 2992, pp. 150, 1117; DX 4593, p. 74);
-All-
L
Z
3
f
7
3
9
io ;
ii
12
13
By the Department of Commerce, Office of Administration
Domestic International Business (360/40), for "Statistical Programs",
"Economic Analysis" and "Industrial Mobilization" (DX 2992, pp. 157,
1117; DX 4592, p. 46) ;
By the Department of Army, U*.S. Army Munitions Command
(360/40), for "Research, Engineering" (DX 2992, pp. 169, 1120; DX 4593,
p. 126);
By the Department of Air Force, Air Force Logistics Command
(360/40), for "Supply, Inventory Control, Cataloging" (DX 2992, pp. 273,
1120; DX 4593, p. 87) ;
By the Department of Air Force, Aerospace Defense Command
(360/40), for "Telecommunications", "Command and Control",
"Intelligence" and "Tracking" (DX 2992, pp.418, 1120; DX 4593, p. 96);
. . t By the Department of Air Force, Air Force Systems Command
I* ;
i
; (360/40) , for "Research, Engineering" (DX 2992, pp. 428, 1120; DX 4593,
ls .
P- 98);
la
j By the Department of Transportation, FFD Aviation Administra--
17 :j
jtion (360/40), for "Inventory, Supply and Logistics", "Mission
- 3 ;
■Support, Operations" and "Planning, R and D" (DX 2992, pp. 1039, 1154;
19 : !
JDX 4593, p. 166) ;
20 ]
i By the Office of Economic Opportunity, Office of the
21 -I
.{Comptroller (360/40), for "Payroll Accounting", "Personnel Accounting"
22. '- .
jand "Research and Development" (DX 2992, pp. 1015, 1146; DX 4593,
23
|p. 191);
i*
15
By the Securities and Exchange Commission, Office of Data
-A12-
1 ij Processing (360/40), for "Mass Information Storage and Retrieval",
2 ;! "Statistical and Economic Analyses" and "Administrative Processing
3
9
10
11
12
(Personnel, Payroll, etc.)" (DX 2992, pp. 1025, 1151? DX 4593, p. 191)
By the Veterans Administration, Department of Data Manage—
- ij ment (360/40), for "Loan Guaranty" and "Facility Planning and
6 !| Construction" (DX 2992, pp. 1073, 1158; DX 4593, p. 195);
t
_ :| By the Veterans Administration, Department of Data Manage-
ment (360/40), for "Patient Care" (DX 2992, pp. 1078, 1158; DX 4593,
p. 196);
By the Department of Navy, Director of Naval Laboratories
(360/40), for "Laboratory Support Systems" (DX 2992, pp. 685, 1123-25;
DX 4593, p. 143) ;
<\ By the Department of Navy, Air Systems Command (360/40) ,
13 j
';! for "Air Logistics Support Systems" (DX 2992, pp. 565, 1123-25;
14 !j
ij DX 4593, d. 139) ;
15 ij
!i By the Federal Deposit Insurance Corporation, Division of
15 : '
:i Research (360/40), for "Economic Research", "Fiscal Accounting" and
17
18
19
20
21
22
23
24
25
"Bank Statistics" (DX 2992, pp. 823, 1130; DX 4593, p. 171);
By the Government Printing Office, Assistant Public Printer j
(360/40), for "Inventory Accounting and Control" and "Electronic
Printing" (DX 2992, pp. 834, 1135; DX 4593, p. 171);
I
i
By the Department of Health, Education and Welfare, Food !
i
and Drug Administration (360/40), for "Disease Prevention and Control"!
i
(DX 2992, pp. 855, 1137; DX 4593, p. 173); j
i
l
By the Defense Nuclear Agency, Headquarters, Field Command
-A13-
1
2
3
4*
3
7
8
9
10
LI
rz
13
14
IS
15 ,.
17 i|
ra I
~ : !
* \
\
•i
(360/40), for "Accounting and Finance", "Nuclear Weapons Materiel Con-
trol", "Stockpile Management" and "Test Command" (DX 2992, pp. 546,
1121; DX 4593, p. 133) ;
By the Department of Navy, Marine Corps (360/40) , for
"Manpower Management System" (DX 2992, pp. 615, 1123-25; DX 4593,
p. 159);
By the Department of Navy, Pacific Commander-in-Chief
(360/40), for "Intelligence Data Handling System" (DX 2992, pp. 753,
1123-25; DX 4593, p. 163);
By the Air Force Aeromed Installation (3 60/40) to simulate
bombing equations, radar signal acquisition and airborne computers
(DX 5640, Mayer, p. 34);
By the Army (van-mounted 360/40s) for maintaining a running
account in the field of supply and demand of field support services
(Wright, Tr. 13394-95; DX 913);
By the U.S. Army Strategic Communications Command (360/40)
for message switching (Wright, Tr. 13412-13) ;
By the Atomic Energy Commission, Chicago Office (360/44),
for "Material Management", "Financial Management" and "Scientific and
Engineering" (DX 2992, pp. 84, 1113; DX 4593, p. 60);
By the Department of Air Force, Air Force Systems Command
(360/44), for "Research, Engineering" (DX 2992, pp. 284, 1120);
££
By the Department of Air Force, Strategic Air Command
I (360/44) for "Command and Control" (DX 2992, pp. 319, 1120? DX 4593,
.! p. 80)';
-A14-
L
I
I
4-
J
/
3
a
3 '
By the National Aeronautics and Space Administration, Office
of Manned Space Flight (360/44), for "Simulation" (DX 2992, pp. 984,
1144; DX 4593, p. 188) ;
By the National Aeronautics and Space Administration,
Advanced Research' and Technical Office (360/44) , for "Test Data.
Acquisition" (DX 2992, pp. 904, 1144; DX 4593, p. 177);
By the National Aeronautics and Space Administration, Flight
Research Center, Edwards Air Force Base, California (360/50) , for
"Scientific", "Engineering", "Data Reduction" and "Business Commercial"
(DX 2992, pp. 905, 1144) ;
By the Railroad Retirement Board, Data Processing and
Accounts Bureau (360/50), for "Research and Actuarial Services",
"Process of Unemployment and Sickness Benefits" (DX 2992, pp. 1021,
1149; DX 4593, p. 191) ;
By the Tennessee Valley Authority, Computing Center Branch
(360/50), for "Resource Development and Management", "Power Supply and
Use", "Fertilizer and Munitions Development" and "Personnel Management".
(DX 2992, pp. 1068, 1156; DX 4593, p. 194);
By the Department of Transportation, Federal Highway Adminis-
tration (360/50), for "Inventory, Supply and Logistics", "Planning,
R and D" and "Mission Support, Operations" (DX 2992, pp. 1050, 1154;
DX 4593, p. 170) ;
By NASA's Flight Research Center in Edwards, California
23 '!
J (360/50), for reduction and analysis of flight data, scientific
?A ■!
* J theoretical calculations and administrative data processing (DX 5 308,
:« I
- ; !p. 1);
:| -Ai5-
■i
.6
.7 ;
,3 :
it
i
.i
7", .1
~ J
: i
L
2
3
4j
By NASA's Kennedy Space Center (360/50) for real time
inventory management, integrated launch, vehicle modification status,
payroll and remote file inquiry (DX 5256, pp. 6, 63);
By the U.S» Coast Guard and Geodetic Survey. Office (360/50)
for developing aeronautical charts , analyzing satellite data , provid-
ing tidal data, locating earthquakes and assisting in geomagnetic
7 ';{ studies (Wright, Tr. 13410-12; DX 13678, p. 9);
By duPont's Savannah River Laboratory Plant (360/50) for
neutron the realization and reactor kinetics (H. Brown, Tr. 83244-49);
By the Department of Navy, Commander-in-Chief Pacific Fleet
(360/50), for "Intelligence Data Handling System" and "CINCPAC Support
Information System" (DX 2992, pp. 752, 1123-25; DX 4593, p. 162);
By the Department of Navy, Commander-in-Chief Pacific Fleet
(360/50), for "Material Management Information System" (DX 2992,
pp. 714, 1123-25; DX 4593, p. 148);
By the Department of Navy, Facilities Engineering Command
(360/50), for "Ordnance Support Systems" (DX 2992, pp. 717, 1123-25;
DX 4593, p. 148) ;
By the Department of Navy, Air Systems Command (360/50) ,
for "Air Logistics Support Systems" (DX 2992, pp. 658, 1123-25;
DX 4593, p. 138) ;
By the Federal Deposit Insurance Corporation,
Division of Research (360/50), for "Economic Research",
I !
By the Government Printing Office, Assistant Public Printer
(360/50), for "Payroll, Earnings and Leave Accounting", "Electronic
Printing" and "Inventory Accounting and Control" (DX 2992, pp. 834,
L I 1135; DX 4593, p. 171);
By the Department of Health, Education and Welfare, Food
^ | ; and Drug Administration (360/50), for "Disease Prevention and Control"
and "Consumer Protection" (DX 2992, pp. 855, 1137; DX 4593, p. 173);
By the Atomic Energy Commission, Albuquerque Office (360/50) ,
for "Facilities Management", "Operations Control and Support" and
"Scientific and Engineering" (DX 2992, pp. 35, 1113; DX 4593, p. 46);
By the Department of Army, White Sands Missile Range, New
Mexico (360/50), for "Research, Engineering" (DX 2992, pp. 177, 1120);
By the Defense Communication Agency, NMCS Support Center
(360/50), for "Gaming, Modeling, and Systems Development", "Command
and Control" and "Damage Assessment" (DX 2992, pp. 551, 1122; DX
4593, p. 133) ;
By the Department of Air Force, Sacramento Air Material
Area, McClellan Air Force Base, California (360/50), for "Personnel"
8 jj
,and "Education" (DX 2992, pp. 273, 1120);
9 . . i
| By the Department of Air Force, Aeronautical Systems Division, j
fl i ■ !
Wright Patterson Air Force Base, Ohio (360/50) , for "Research, Engineer- \
1 •
!ing" (DX 2992, pp. 282, 1120);
~ 1 I
:j By the Department of Air Force, Pacific Air Force (36 0/50) , !
2 | I
•I for "Command and Control" (DX 2992, pp. 456, 1120; DX 4593, p. 105); j
i By the Department of Air Force, Strategic Air Command j
"1(360/50), for "Intelligence" (DX 2992, pp. 322, 1120; DX 4593, p. 80); j
J
r
■
y i
i
1
m
6 (
7 I
i
ii -A17-
1
2
2.
4*
6
7
3
9-
10
II
12
13
14
15
is ;
17
is
19 ; l
By the Atomic Energy Commission, Savannah River Office
(360/65), for "Material Management", "Financial Management" and
"Scientific and Engineering" (DX 2992, pp. 115, 1113; DX 4593, p. 71);
By the Department of Army, Safeguard, Whippany, N.J.
(360/65), for "Research, Engineering" (DX 2992, pp. 165, 1120);
By the Department of Air Force Ogden Air Material Area,
Ogden, Utah (360/65), for "Payroll, Benefits", "Procurement, Contract
Administration" and "Law Enforcement" (DX 2992, pp. 275, 1120);
By the Department of Air Force, Air Force Systems Command
(360/65), for "Research, Engineering" (DX 2992, pp. 438, 1120; DX
4592, p. 67);
By the Department of Air Force, Aerospace Defense Command
j (360/65), for "Command and Control" (DX 2992, pp. 418, 1120; DX 4593,
p. 96);
By the Department of Air Force, Air Force Systems Command
(360/65), for "Intelligence" (DX 2992, pp. 428, 1120; DX 4593, p. .97).;
By the Defense Communications Agency, NMCS Support Center
(360/65), for "Gaming, Modeling, and Systems Development", "Command
and Control" and "Damage Assessment" (DX 2992, pp. 551, 1122; DX 4593,
;|P- 133);
20 i
By the Department of Navy, Marine Corps Automated Service
21 •'!
jCenter, Kansas City, Missouri (360/65), for "Manpower Management
ISystem" and "Personnel Accounting System" • (DX 2992, pp. 628, 1123-25);
23 I
' By the Department of Navy, Naval Air Development Center,
** -iwarminster, Pennsylvania (360/65), for "Laboratory Support Systems"
25 j
(DX 2992, pp. 568, 1123-25); I
i
-A18-
L i By the Department of Interior, Geological Survey (360/65) ,
2 l for "Recreation Use and Preservation" (DX 2992, pp. 877, 1140;
DX 4593, p. 174) ;
By the Department of Labor, Departmental Data Processing
Center (360/65) , for "Accounting and Payroll Services" and for
"Statistical Data Gathering" (DX 2992, pp. 883, 1142; DX 4593, p. 175)
By the National Aeronautics and Space Administration,
Goddard Space Flight Center (360/65) , for "Scientific" and "Engineer-
ing" (DX 2992, pp. 908, 1144);
By the National Aeronautics and Space Administration,
Goddard Space Flight Center (360/65), for "Data Reduction" (DX 2992,
pp. 907, 1144);
By the National Aeronautics and Space Administration,
>, j| Johnson Space Center (360/65), for "Simulation" (DX 2992, pp. 983,
, s \ 984, 1144) ;
By the Tennessee Valley Authority, Computing Center Branch
(360/65), for "Power Supply and Use", for "Fertilizer and Munitions
Development" and for "Employee Health and Safety" (DX 2992, pp. 106 8,
1156; DX 4593, p. 194) ;
4-
a
7
3
9
a
L6 !
IS
:a1
j By NASA's Johnson Space Center (360/65) for Sky lab
i simulation (DX 7536, Woodling, pp. 23-24) ;
21 '*.
j By the Navy Computer Sciences Department in San Diego
22 ;i
I (360/65) for processing complex scientific and management type data
| and for time sharing (DX 5100, pp. 17, 28);
Z- !
| By the Air Force Eastern Test Range (3 60/6 5) for mechanized
-A19-
range scheduling, radar data reduction, trajectory measurement, optical
infrared system data reduction (DX 5023, pp. 1-4);
By the California Institute of Technology's Jet Propulsion
• Laboratory (360/75) for real time mission control, simula-
tion and real time telemetry (DX 5296, pp. 4, 6, 7);
By the Naval Electronics Laboratory Center in San Diego
(360/65) for interactive time sharing (DX 4334, pp. 1, 5) ;
By the Department of Air Force, Air Force Systems Command
(360/67), for "Telecommunications" and "Command and Control" (DX 2992,
pp. 451, 1120, DX 4593, p. 103);
By the Defense Communications Agency, NMCS Support Center
(360/67), for "Command and Control" and "Damage Assessment" (DX 2992,
14- ;'
By the Department of Navy, Post Graduate School (3 60/67) ,
. ! for "Management Information System for Education and Training" (DX
U3
2992, pp. 588, 1123-25) ;
By the National Aeronautics and Space Administration,
Ames Research Center (360/67), for "Scientific" and "Business-
IS
17
i Commercial" (DX 2992, pp. 888, 1144);
I By the Atomic Energy Commission, Idaho Office (360/7 5) , for
20 ]
'l "Material Management", "Financial Management", "Personnel Management"
21 4 .
"j and "Operations" (DX 2992, pp. 77, 1113; DX 4593, p. 58);
22 ;!
;j By the National Aeronautics and Space Administration,
23 ;
I Goddard Space Flight Center (360/75), for "Scientific", "Engineering"
2* 1
j and "Mission Control" (DX 2992, pp. 907, 908, 1144);
25 !
j
i -A20-
•i
■i
r
•I
'i
it
LI
By the National Aeronautics and Space Administration, Jet
Propulsion Laboratory (360/75), for "Data Reduction" (DX 2992, pp.
I I 946, 1144);
i-
i .
f r
z.
7
S
a
i
i
A
«*
.5 .1
■7\
3 ;
"* .1
I
LI =!
\
■2. ^
i
.- ii
'i
-A21-
1 35. The System/360 Commitment . System/360 was a "fantastip
2 • undertaking" involving "fantastic risks". (cary, Tr. 101359; see
3 j also Brooks, Tr. 22868; Case, Tr. 73561; Evans, Tr. 101126.) 360 was
^ : "vastly different" from anything IBM had previously undertaken in
5 : terms of "magnitude, complexity and functional characteristics", and
6 • was "fundamentally new and different" compared to competitors' EDP
7 offerings as well. (Knaplund, Tr. 90515; Evans, Tr. 101126; PX 1092
8 p. 1; DX 117'2, pp. 1-2.) It was clear from the outset that no half-
9 • way measures would suffice to carry out the SPREAD Committee's plans— f
10 | and non was taken. IBM committed more "skill and energy" and
11 "corporate resources" to the successful implementation of System/360
12 than to any previous undertaking in its history. (PX 1900, p. 4.)
Virtually the whole IBM's EDP operations were involved in
the development and manufacture of System/360. The scope and magni-
tude of the undertaking required a worldwide, interdivisional effort
on IBM's part. "From its inception, System 360 was designed, dev-
17 i eloped and tested for worldwide use, and was in fact used worldwide".
18 I (McCarter, Tr. 88377; DX 1404A, p. 8 (App. A to JX 38).)*
19
20 *The 360/30 was developed in Endicott and was manufactured in
Endicott, Sindelf ingen, Germany, and Mainz, Germany. (Dunlop, Tr.
21 93647.) The 360/40 was developed in Hursley, England, and manufac-
tured in Poughkeepsie, Essones, France, and Montpellier, France.
22 (Id.; Hughes, Tr. 33921-22.) The 360/50 was developed in Poughkeepsi^
ancT manufactured (assembled) in Poughkeepsie, Essonnes, and Mont-
pellier. (Dunlop, Tr. 9 3649.) The 360/20 was developed in Boeblin-
gen, Germany and manufactured (assembled) in Sindelf ingen, Vimercate,
Italy, San Jose, and Boca Ratan. (Id.; Hughes, Tr. 71942-43.) System
360' s SLT circuit packaging was designed in Endicott and East Fishkii:.,
and manufacturered in East Fishkill, Endicott, Essones and Sindel-
f ingen. (Dunlop, Tr. 9 3649-50.) The 2401 tape subsystem was dev-
eloped in Poughkeepsie, and manufactured (assembled) in Poughkeepsie,
Essones, Montpellier and Boulder. (Dunlop, Tr.
-341-
13
14
15
16
23
24
25
7
3
1 !
Within IBM, it was recognized that achievement of SPREAD'S
recommendations would require "great effort" to "control and
coordinate the work of several divisions and that of the IBM World Trade
•Corporation". (Knaplund, Tr. 90470-71.) At the time of SPREAD there
were 15-20 engineering groups generating processor products in IBM.
(DX 1404A, p. 7 (App.. A to JX 38).) These groups resided in four
principal areas — DSD, GPD, FSD (Federal Systems Division) and WTC (World
Trade Corporation). (DX 1404A, p. 49 (App. A to JX 38).) If a single
compatible line of processors was to be achieved, design control had to
be centralized in a single location.* Accordingly, the SPREAD Committee
recommended the establishment of a systems architecture group that would
be charged with formalizing the design objectives for NPL and providing
it
illogical specifications for the hardware and software. (DX 1404A, p. 49
• (App. A to JX 38).) Such a group — the NPL Architecture Committee — was
formed in early 1962, and served in the role of "advisor" to the various
A
.0
.7 ■ 93650.) The 1403N1 printer was developed in Endicott and manufactured
•in Endicott, Raleigh, Sindelfingen and Vallingby, Sweden. (Dunlop,
p S ;Tr. 93650-51.) The 2311 was developed in San Jose and manufactured in
uSan Jose and Sindelfingen. (Dunlop, Tr. 93651.) The 2671 paper tape
[g 'recorder was developed in LaGaude, France, and manufactured in Essonnes
:jand Montpellier. (Dunlop, Tr. 93651.)
:J * Centralized control of worldwide development efforts made good
2i [sense from another standpoint as well. The SPREAD Report projected
;;a very rapid increase in the growth of computer usage outside the United
22 i States during the 1960s; whereas the average domestic growth rate was
(projected to be 15%, the foreign rate was projected at 37%. (DX 1404A,
?- ! |p. 11 (App. A to JX 38).) So large an element of demand obviously
""" jcould not be ignored in the development of new products, and the
2 4 ijCommittee recommended that the needs of users worldwide be taken into
(account in all phases of NPL development. (DX 1404A, p. 49 (App. A
„- : |to JX 38) .)
£■3 i
J -342-
!
L
2
2
4-1
f.
7.
8j
5 i
io i
a!
i
12. '
is :
i
L4.i
^ i
NPL engineering groups. (Case, Tr. 74487-88, 74492-9 .) They held
"dozens if not a hundred or more meetings" relating to NPL. (Case, Tr.
74469.)
On the. manufacturing side, too, a number of disciplines were
: imposed to assure that there were no major discrepancies among the
products produced on either side of the Atlantic. IBM's plants
worked "very closely" together to develop "worldwide manufacture
plans" and employee training plans. (Dunlop, Tr. 93651-52.) IBM also
introduced, for the first time with System/360, the concept of
"single engineering control". (Dunlop, Tr. 93641, 93646.) Under this
concept any laboratory responsible for designing a part, component or
product was also responsible for releasing that design to all the
plants, worldwide, that were going to manufacture that part, component
,or product. (Dunlop, Tr. 93641.) By introducing this concept, IBM
was able to:
(a) achieve a "high level of confidence" that all parts,
1i»
C if
■| wherever in the world produced, would perform in a comparable
17 \
,'| fashion;
IS
\ (b) achieve the ability to exchange parts or assemblies
IS' :
•J or products among manufacturing locations in times of tech-
2G ]
j nological difficulty or great demand;
21- If •
i (c) avoid duplication of engineering effort, since there
j was no need to design the same product or component twice in
j two different places. (Dunlop, Tt. 93642-43, 93645.)
24 !
j Apart from the need to impose new disciplines, it was
25 j " ■
-343-
3-
<•
ri
a
.a
■21
14- i
apparent that a "substantial" segment of IBM's "new product develop-
ment resources in the electronic data processing (EDP) area"
would be required to announce the New Product Line in the first quarter
^|of 1964. (Knaplund, Tr. 90471.) Brooks testified that the original
estimate for 360 programming was between $100 and $200 million.
(Tr. 22706.) That estimate was exceeded by better than $25 million.
(Id. ) Brooks' staff in DSD alone grew from "20 or 30" in June 1961
to "several hundred" by February 1964. (Brooks, Tr. 22669.) A presenta-
tion made to IBM Chairman, T. J. Watson, Jr., in November 1964 showed
that IBM's annual research and development expenditures rose from
approximately $175 million per year in 1961 to $275 million per year in
1964. (PX 6671, p. 6.)*
More investment still was needed to meet the requirements
for SLT components. The 1961 decision to manufacture SLT in-house
required a rapid buildup in manufacturing facilities and resources .
(Knaplund, Tr. 90546; E. Bloch, Tr. 91562.) To meet the projected volumes
» i
lifor 360, IBM had to become "in a very short time, the largest component.
| manufacturer in the world". (PX 1900, p. 9.) In 1961, IBM established
;ja Components Division to "focus all of its resources in terms of both
; manufacturing and development on that goal of making SLT components."
'6
17 ;
IS
19
*:j
.1
21 ;i
: - * It is interesting to note that, in the 1959-64 period, IBM's
2- ^research and development (R&D) expenses were not only absolutely highe:
;jthan some of its major competitors (Burroughs, NCR, Sperry Rand and
-- (CDC) , but were more than double the expenditures as a percentage of
^ jrevenue for Burroughs or NCR or Sperry Rand. Each of their ratios
j 4 :jof R&D to revenue remained about level over that period. Among the
"""" jfour, only CDC, which was developing the highly successful 66 00,
0jC -(showed an increasing R&D to revenue ratio. (PX 66 71, po. 5-6.)
I
I -344-
Li
2 ;
i
i
3
i
5
c
7
a
9
LQ
II
12.
13
15
15
17
IS
(E. Bloch, Tr. 91562, 91891-92.) In 1963, the Components Division
opened a new plant in East Fishkill, New York as a manufacturing
development site for System/360 components. (E. Bloch, Tr. 91563,
. 91891-92.) Prior to the 360 announcement, IBM hired "a large number of
people" and "started to build additional buildings" in order to meet the
anticipated SLT requirements. In addition, IBM's Endicott location was
enlarged to help produce packages for mounting SLT modules (E. Bloch,
Tr. 91892) and part of a plant in Essonnes, France, was converted into a
"component facility" to help meet "worldwide requirementC s ] " . (E. Bloch,
Tr. 91893, 91563-64.)
Perhaps as significant as the magnitude of IBM's investment
in 360 was the fact that all of those resources were being put into a
single project: IBM was "putting a lot of eggs in one basket ...
and the success of the company was in many ways to be determined by
the success of that one project". (Case, Tr. 73561; Evans, Tr.
101128.) If 360 were rejected by customers there would be few alterna-
tives around for IBM to offer and none that was thoroughly funded or
'covered a very large part of the product line. (Case, Tr. 73 562.)
IjThus, once the die had been cast and the decision made to go forward
19 ;• ■ ,
•;with the SPREAD Committee's recommendations, IBM's fortunes became
:i " inextricably tied up with the NPL project". (Case, Tr. 73562.) Within \
2i i •
:IBM and without, the 360 project came to be known as the "you bet \
22 "• !
'your company" venture. (Friedman, Tr. 50378; Case, Tr. 73561-62; j
| Evans, Tr. 101126.) If that venture had failed, IBM would have
2.L 'I
jbecome a "radically different company, if even in tne computer
~ :|business". (Evans, Tr. 101128.)
-345-
3
.Q
Despite the risk, IBM decided to develop the 3 60 line
because " [Rethought that the System/360 development was the best way
to more rapidly grow the market, more rapidly expand demand for our
products". (Case, Tr. 73606.) It was the sort of risk that IBM was
forced to take by competition if it was to succeeds Seemingly safer
alternatives to 360 continued to be advanced within IBM right up to
j ,j the time that 360 was announced, (See, e.g., Case Tr. 73589-92; PX 1074;
PX 1090.) As they had rejected the 8000 series, IBM management rejected
those alternatives because they would not have given IBM the kind of
long range solutions that it needed in the competitive environment of
the day. (Evans, Tr. 101277; see also DX 4806.)
System/360 represented a price/performance improvement over
:i
.^ \ IBM's existing eauipment which Learson described as "a price reduction
^ :!
of 30-50%." (DX 1525.) Within IBM, it was recognized that no "single
announcement" had ever "obsoleted so much existing equipment at one
time". (PX 1099A.) IBM was forced to make such an announcement.
[The SPREAD Committee had set as an objective the creation of a plan
that would "optimize the conflicting demands" of "market need" on the
is ;|
lione hand and "impact on present installed processors" on the other
® il
: (DX 1404A, p. 7 (App. A to JX 38)) — but IBM had to impact its own line
2G
:jor stand by and watch others do so.
21 -\ .
;j In an effort to blunt the impact of System/360 on IBM's
22. ; i
{existing product line, IBM Treasurer K. N. Davis recommended that 36 be
23
{offered for sale only. Davis made the suggestion because technology and
Z± \
jprice/performance were "changing and improving so rapidly" that
-?;,
-346-
he believed it might be in IBM's interest to transfer to customers
some of the risk of technological obsolescence. In addition, System/
360' s price/performance on a rental basis was so superior to existing
. IBM systems on rent that customers would rapidly displace those
systems with 360s. (Knaplund, Tr. 90511-12.) The recommendation
was rejected because "IBM had to continue to offer a rental option
in order to remain competitive": competitors offered that option and
customers found it desirable. (Knaplund, Tr. 90512-13.) In this
respect, IBM's experience was no different from its competitors. For
example, McCollister testified that RCA offered its systems on a lease
basis because the customers insisted upon it and because all other
manufacturers in the industry offered it. (McCollister, Tr. 9292-300;
see also Palevsky, Tr. 3145-46; Spangle, Tr. 5531; Oelman, Tr. 6160.)
Indeed, customers as well as IBM could perceive that technology was
changing and would not have been willing to accept the risk of obsoles-
cence. Competition ensured that they did not have to do so.
As Withington agreed, IBM had to introduce a product line
comparable in performance and function to System/360 if it wanted to
stay in business because its existing line would have become "obsolete"
and unmarketable. (Tr. 56524, 56539.) Thus, IBM Vice President and
2G I
•I Group Executive Learson wrote to C. J. 3ashe, Manager of Technical
i
21 \
•Development, GPD, and T. C. Papes, Manager of Systems Development, GPD,
*— \ in July 196 3:
L
2
3
4-
£
7
3
a
10 i
!
11 !
12 i
I
13 ;
i
14*;
i
15 ;
i
15 ;
17 ;
IS
19
24 I
?«; i
"The 101 [announced as the System/360 Model 30] must be
engineered and planned to impact solidly the 14 01.
"I know your reluctance to do this, but corporate policy
is that you do it. It is obvious that in 1967 the 1401 will
-347-
be as dead as a Dodo bird.E*] Let's stop fighting this."
(DX 1406.)
Hughes testified that this letter was passed down to him
i through the management chain to emphasize the importance of the 36 0/
3
I 30 program and the company's policy with respect to that program. He
; understood that the 360/30 would make the 1400 family obsolete — and
5 ;
! had to do so. (Tr. 33965-66, 33972-73.) Despite the fact that by
7 | 1964 IBM had shipped thousands of 14 01 svstems, of which 75-80
7 !l
percent were still owned by IBM and on lease to customers, it was pert
ceived that " [i]f we didn't obsolete it and replace it, someone else
would". (Hughes, Tr. 33962-63, 33965; see also Tr. 34062.) That
same view was echoed in a letter written by a Staff Vice President to
the President of Southern Railway in April 1964, recommending the
acquisition of 360/30s to replace Southern's 140^ s:
8
9
10
11
12
13
14
15
16 !
i
17 !
18 !
19
20
21
22
23
24
25
"This will reduce the IBM rentals by $4,000 a month in
Atlanta. There is also a good possibility that we will be
able to eliminate the 1401 computer in Washington, using
computers in Atlanta by tape to tape control from Washington.
This would also save us $4,000 to $5,000 per month rental in
Washington. Prices of computers have been coming down while
the computer capacities are being increased tremendously.
If IBM does not bring out new computers at reduced prices,
their competitors take the business."
According to John Jones of Southern Railway who helped draft that
letter, it reflected his view of competition in 1964 — i.e., that if
IBM and others did not bring out new products to meet competition, com-'
petitors would take their business away — the kind of competition which
*By year-end 1966, I3M had installed over 10,000 1401s, far and
away the largest number of any system type that IBM had ever shipped
at that tine. (?X 190 0, p. 7.1
-348-
L
2
3
4*1
5 I
I
/«■ '
r\
8
9
1Q
U
IZ
12
had increased "tremendously" since then. (J. Jones, Tr. 78991-97.)
It was a view that was shared by IBM's competitors as well
"There is no looking backward in our industry [the com-
puter business] as you undoubtedly know. If one stops to
ponder the past and be self-satisfied, the more aggressive
competitors will quickly charge past." (Hindle (DEC), Tr. 7447
DX 517, p. 2j
and
"It was our finding that the life of a family of com-
puters was quite limited . . . and that you did not bring
out a family of products that simply met the price/perform-
ance characteristics of the then existing competition. You
had to bring out something that would exceed the price/
performance of the existing competition because you knew
full well that they were going to be moving ahead of you.
It is a constant leap frogging game." (R. Jones (GE) , Tr. 8867)
One gets "to a point in which the price/performance is
so improved over equipment of days of yore that it is
clear that . . . users are going to move to new equipment,
and either [one is] going to provide that new equipment
or [one's] competitors are going to provide it". (R. Bloch
(Honeywell/GE) ,Tr. 7761-62; see also Hindle (DEC), Tr. 7448;
R. Jones (GE) , Tr. 8865; Hangen (NCR), Tr. 10423-24, 10431;
Currie (XDS) Tr. 15175-76; Brooks, Tr. 22705, 22795-96; Withincton,
IS I! Tr. 56560, 56565; DX 426, pp. 7-8.)
17 ; As we have already discussed, it was the recognition that
IS ; competitors would supplant IBM's installed base if IBM took no action,
19 i!as reflected in the SPREAD Report's "product survival charts", which j
:j I
2G ihad triggered the NPL project to begin with. The SPREAD Committee's j
*" 1 I
21 [prediction that IBM's highly successful second generation line would be j
22 "[superseded by competition starting in about 1965, turned out to be |
i !
23 ''accurate as to substance, but overly optimistic as to time. As j
!
24 'jWithington testified, the industry was in a state of "technological fer- !
I '
2= "ment during the period 1956 through 1964", with "new technologies . . . \
■i •
j -349-
i i
'i I
! !
,t I
/
3
3
a
2
3
new types of components . . . [and] significant software products . . .
being invented and employed at a rapid rate" and new models of computer
systems superseding older computer systems at a "rapid rate" and
achieving "relatively rapid success in the marketplace". (Tr. 56459-60.;
In 1963 and early 1964, the "leapfrogging" which was "characteristic"
of the computer industry (R. Jones, Tr. 8846) had occurred. In July
of 1963 Learson could say that "in 1967 the 1401 will be as dead as a
Dodo bird" (DX 1406) because it was already being surpassed by newer
models of computer systems.
Indeed, at the highest level within IBM there was concern
that the System/360 might not be enough of an improvement to recover
its costs. Thus Watson, writing to Learson in June 1963, stated
concerning the New Product Line:
"I think it important to note, however, since we seem
to have suffered for a few months or even years because our
machines predated the effective competitive machines now in
the marketplace, that we now make these [System/360] machines
2 ij good enough so they will not be just equal to competition,
for I am sure that once they are announced our competitors
will immediately try to better them. This is all to the good
and I am for competition, but I want our new line to last
long enough so we do not go into the red." (DX 4806.)
,a i
.7
.3
i
{Similarly, writing in November 1963 to a group of IBM executives,
.} Watson said:
•0 .1
•I
?7 !
"There is a great deal of running about and extra effort
being expended in all areas of the IBM company now because
-I once again we have allowed ourselves to become somewhat non-
12 '] competitive without recognizing one simple obvious fact. In
,| bringing new machines and devices to the marketplace, our
23 j competitors in today's market are simply not going to stand
j still. We should recognize that in every area, they will
Z4. '; take the best we have and immediately start working in a
j tough, hard-minded fashion to produce something better.
* s '
> •■
I "We find ourselves in our present position because we
! seem to assume our competitors will stand still in certain
-350-
L
2
3
4-
5 If
a >
7 1
3
9
ia
ii
12
13
14-
15
La
17
areas after we announce a superior product ....
"I believe that whenever we make a new machine
announcement, we should set up a future date at which point
we can reasonably assume that a competitor's article of
greater capability will be announced. We should then
target our own development program to produce a better
machine on or before that date*" (PX 1077, pp. 1-2.)
Charts prepared by DSD Market Evaluation Manager, J. C.
Wick, comparing the price/performance of the New Product Line to com-
petitive products in February 1964, showed that 360' s price/performance
was superior to that of recently announced machines from RCA,
Burroughs, CDC, Honeywell, Univac and GE, but also showed quite clearly
that those competitive machines had a price/performance advantage over
the earlier announced IBM machines of the 1400 and 7000 series. (PX
1099A, pp. R2-R3.) We discuss some of the competitive announcements
which created this situation in the histories of these competitors
during the early 1960s. However, some of the announcements merit
particular attention here.
In October 1963, DSD President G. F. Kennard wrote to T. J.
Watson, Jr., and A. L. Williams: "RCA has recently announced the
.- =3301. . . . Initial performance specifications indicate that the 3301
- S 1
ilhas about 50 percent better processing capabilities than the IBM 7010"
2- I at a comparable price. (PX 2952.) In November, 1963, it was
i
7t 'reported within IBM that GE was discussing in public a new series of
^ machines planned for announcement before the end of the year. "In one
•[case GE stated, system cost would be approximately the same as the
22 I
IBM 1410 but would be 40% faster." (PX 36 24, d. 4.) GE announced the
24 i
J
,i
•?« '<
-^ i
!
•j -351-
■i
:i
400 Series in December 1963,* and at the same press conference revealed
the future availability of its 600 family.** (Weil, Tr. 7181; DX 488;
DX 490..) The 400 series offered a 1401 simulator which permitted IBM
1401 programs to be run on or converted "easily" to the 400. It was
aimed at 1401 users* (Weil, Tr. 7031-34.)
'r'l The CDC 6600, which CDC began discussing with customers
i|
l jj before announcement (Norris, Tr. 5937-38) in 1962 (JX 10, u 4), caused
IBM Chairman Watson to ask "why we have lost our industry leadership
position by letting someone else offer the world's most powerful
computer". (PX 1045.) CDC's 3600, which had been announced in May
1962, was viewed within IBM as "technically superior to the 7094".
(PX 1026A.) By April 1963, 0. M. Scott, IBM Vice President and Group
Executive, was reporting to Watson and others that "3600-type com-
petition" was creating a "serious situation" and that such competition
(from CDC's 3600 and 6600 and from Philco ' s 212) was able to offer
7 !
S
.9
'a
I "one-and-a-half to two times the performance of the 7094 at a lower
price". (PX 1025.) Scott added that the 501 (360/70), as planned,
would enable I3M "to favorably compete with the CDC 3600". ( Id. ) On
April 23, 1963, Watson determined to "just sit tight" and stay with
the 501 approach "unless the roof falls in", but wrote that IBM had an
:u :j
, active program in DSD called the "7094 B crime"' which vras sufficiently
LI : !
^advanced to be announced in June 1963. (PX 2807.) Within two weeks,
•2 ;, i
?i | * GE announced the Models 425, 435, 455 and 465. The 455 and 465
iwere never delivered. (Weil, Tr. 7181; DX 4 90, pp. 1-3.)
15 I
| ** The 600 family was actually announced in the summer of 1964
'■'(Weil, Tr. 7197-98; DX 491, p. 1) and was aimed at IBM 7090 and 7094
lusers. (Weil, Tr. 7033-38.)
: !
1 -352-
3
4-
5
^
a
cascading losses to CDC's 3600 caused a reevaluation of that decision,
and Watson asked Scott to advise him when the situtation got "out of
control". (PX 3619.) One week later, Scott reported back that IBM
was repeatedly "being beaten" by CDC's 3600, 6600 and 1604, Philco's
212 and Remington ' Rand ' s 1107. He recommended announcement of the
7094-B' "at the earliest possible date". (PX 3620.) IBM announced
^ .{the 7094 Mod. II on May 16, 1963 (DX 13958 ) r but this extension of the
7090 series still "could not meet either the performance level or the
price of a comparable CDC 3600." (PX 320, p. 15.) As a result, CDC's
success with the 3600 continued unabated. (PX 320, p. 15.) With
virtually all of IBM's development resources tied up on 360, IBM was
simply unable to respond effectively at that time — all of IBM's eggs
were indeed in the 360 basket.* (See Case, Tr. 73589, 73561; Evans,
•|Tr. 101128.) In the meantime, CDC was able to achieve success "by
concentrating on an area of IBM price weakness, and by showing a major
price performance advantage to potential customers". (P.X 320, p. 15.)**
3 '
* I
j
ii !
12.
L4- :
Perhaps most important of all, however, was the announcement
IS ,
17 ;
is ;
xs i
:■{ * At just about this same time CDC's chief development engineer
?g '-for the 6000 Series, Vice President Seymour Cray, at CDC's June 1963
'corporate planning meeting, urged that CDC announce the 6600 and a
2i ^successor in order to "slug" IBM because he speculated that IBM had
'; "rradea mistake in putting all [its] eggs in an integrated circuit
kef. (DX 13526, Forrest, pp. 748-50O
** No competitor was able to offer such an advantage once 360 was
^ (announced. (PX 320, pp. 4-14.)
£Z
-353-
J i
7:
3
9
a
i
2.
of the Honeywell 200 in early December 1963. (McCollister, Tr. 11367;
PX 1Q79; DX 167) This machine offered substantially improved price/
performance over the 1401. (McCollister, Tr. 11237; Knaplund,
Tr. 90475; Evans, Tr. 101188; DX 167.) It also offered a conversion
«
program called the "LIBERATOR" which made the H-200 to a considerable
degree compatible with IBM's 1401. (R. Bloch, Tr. 7605-06; McCollister,
Tr. 11237; Goetz, Tr. 17652; DX 167; DX 488.)
Within IBM the H-200 announcement was viewed as "even more
difficult than we anticipated". (PX 1079.) Within two days of the
announcement, Lear son wrote to T. J. Watson and A. L. Williams that
the 101 (360/30) would have to be announced "as soon as possible"* and
priced at its "lowest projection" in order to be competitive. ( Id. )
IBM's marketing force regarded the H-200 as a real challenge (Evans,
Tr. 101186) and at least one person in I3M called it "the most severe
threat to IBM in our history". (PX 3912.) By February of 1964, the
Sales Division was "reeling from losses" to the Honeywell 200 and
jj "wanting a more competitive answer". (Evans, Tr. 101196.) 3ecause of
lithe H-200, IBM's Data Processing Division continued and intensified
LS \
;jits pressure for the earliest possible announcement of System/360,
19 i
'•{earlier still than even the then-planned mid-March announcement date.
IQ i
:j (Knaplund, Tr. 90475; JX 38, 11 16; PX 1095; DX 2983.)
21 :!
1 As competitive pressure mounted, the debate whether to go
; j forward with 360 as planned or to announce extensions to the existing
I
"" j * The target announcement date at that point in time was March 196 4
- s ) (PX 1079.)
-354-
L
1
3
4-
a
m
a
product lines was rekindled. The latter approach would be safer and
easier: it would not be as "revolutionary" as 360 and would therefore
run a lower risk of user rejection. (See, e.g., Case, Tr. 73590, 73512;
Evans, Tr. 101127.) Moreover, it would not require users to convert
their existing applications programs. In November 1963, IBM's
Corporate Staff advanced the position that "new marketing developments"*
the announcement, in May June 1964 of "several improved current line
systems — such as the 7074X, 7010X and 7094X" . In their scheme of
things the NPL announcement was to be put off for 6 to 12 months.
(PX 1074, pp. 2-3.)
The Honeywell 200 announcement provided perhaps the sharpest
temptation to depart from the System/360 plan. In early 1963, IBM had
?<t ;j a 1401 built out of SLT circuitry to establish the feasibility of using
I SLT in the New Processor Line, ("Hughes, Tr. 33952—53.7 McCarter,
I Tr. 88394; JX 38, 1( 7; DX 4800.) The Honeywell 200 prompted
I sharp debate within IBM whether a new technology (SLT) version of the
| 1401 (called the 140 IS) should be brought out and the 360/30 announce-
ment delayed or cancelled. (Hughes, Tr. 33953-54; Evans, Tr. 101188,
is
17
IS
13
;| 101195.) The chief proponent of this new plan was GPD President John j
'* i
IHaanstra, who had been Chairman of the SPREAD Committee. Haanstra
21 1 . |
il
22 :;
<?- ; !
- 3 j * Thsse new developments included the announcement of competitive
| j processors offering easy conversion to IBM customers and other new com-
*- jpetitive offerings with improved price/performance as well as "the
;« continuing unattractiveness of programming conversion and associated
*5 ..expense to our customers". (PX 10 74, p. 2.)
1 !
;j -355- i
I !
2
3
3
C
7
3
9
.0
12
13
14
believed that the 1401 was a "fundamentally sound" approach to meeting
user needs and that the 360/30 approach was "improper" because it
created the exposure of requiring customers to convert:
"We must have a position which sticks to the 1401 as a
fundamentally sound andproper [sic] method for commercial data
processing. I do not believe that we should in the GP small
machine area imply in any fashion whatsoever that the 1401
approach to problem solving is out of date and that people
must change.
« • » •
"... [I]n the final analysis we must sustain a position
of 1401 as a right programming approach now and into the
future. An approach which implies that we must convert is
basically improper." (PX 3913.)
The Data Processing Division, however, regarded the 1401S as
only a fallback position in the event that the 360/30 was not ready
soon enough or was not good enough:
"The best solution ... is a 101-H machine with a
competitive price to the H-200 and a performance equal to
or greater than the H-200, ready for announcement by mid-
February. . . . This system would not only compete head
r g ; on with the H-200 but offer the customer the opportunity
to grow in the NPL line, which is the direction we want
them to take.
j "The 140 IS machine, which has been discussed, is a
i second choice to the system described above and has been
! supported by us only because we have not received a
commitment that the 101-H machine could achieve the per-
formance desired or meet an early announcement schedule."
(PX 1090.)
~ ! Evans was sure that it was a mistake to produce the 140 IS instead of
4
^ | the 360/30, and that it would not make sense to do both. As earlv as
r- l
- 3 j September 1^63 he had inveighed against "continual competition with
-- ! temporary machines" because they would "only dilute [IBM's] already
25 | overcommitted resources and abilitv to meet the NPL challenge". (DX
i
1 -356-
L
2
4-
3
a i
7{
3
9
ia
u
12
13
15
IS
2983.) In his view, if the 1401S had proceeded, it would have "delayed
if not killed" the 360/30 and "wreaked havoc with the costs of the rest
of the System/360 line". (Evans, Tr. 101195-96.) In addition, Evans
regarded a decision to produce the 1401s as relegating the NPL more to
the scientific area and signalling "a discrete scientific line/ probably
along the 7090 philosophy particularly if competition does the H 200
type of thing to the 7090 family". He felt this would erode the basis
for NPL and lead to a processor policy of "discrete 1400-type commercial,
discrete 7090-type scientific, plus various custom units for new
application areas", as "the inevitable conclusion". (Evans, Tr. 101275-
76; PX 6668 (DX 14514).) As we shall see below (pp. 379-81), GE
ijwas in fact attempting to do "the H 200 type of thing to the 7090 family.
Evans was right .
Although contingency plans were laid for a possible February
1964 announcement (PX 6202) , IBM decided not to proceed with the
1401S. Evans testified that the 1401S was ultimately rejected
: "[b]ecause the evaluations and conclusions of senior management
17 ; were that it was not an advanced system that would solve the
■\ applications of the future as we then saw them — that ... it
IS •! was a machine that would not have long life and would not be
:j competitive for more than a short period, and that the 36
13 j family plan with all of its advanced features and functions
:( and capability and the unusual power it brought the users was
2Q'! a substantially better plan". (Tr. 101277.)
2i 'in short, the 360/30 was expected to be "a better overall performing
22 \ system than the 1401 had been or could have been, had we extended its
22 [life". (Hughes, Tr. 33953-54.)
\
2* \ a. Preparation for Announcement. It was clear by the end
.i
-- lof 196 3 that announcement of Svstem/36 was required for IBM to remain
i
:1
■! -357-
L
I
I
4.
competitive. (Knaplund, Tr. 90475.) We have already discussed how,
beginning in 1961, IBM began applying massive resources to the NPL
project. Evans testified that the "whole 360 program had been on a
. crash basis . . . since almost inception" and that by the latter part
„
of 1963 it had become an enormous program with its own inertia".
„ !' (Tr. 101190, 101198^99.) In December of 1963, development of the line
was "on or ahead of the schedule called for two years earlier in the
SPREAD report" (Knaplund, Tr. 90477),* and "two of the prime movers of
the project, Evans and Brooks, were recommending announcement of the
entire family in the first part of 1964.**
/
3
.0
1 i
* A PERT chart (DX 1405), prepared by Ernest Hughes in October 1962,
laid out the job to be done in order to accomplish the Model 30
program. (Hughes, Tr. 33933-34, 33947.) The chart showed that the
Model 30 would be ready for first customer shipment on August 1, 1965,
if the sequence of events identified on the chart were "successfully
.4 j' completed". (DX 1405; Hughes, Tr. 33947.) According to Hughes, all
of those tasks were completed "close" to the dates projected for their
completion back in 1962, and the first 360/30 was actually shipped in
June 1965. (Tr. 33947-49; see also JX 38, 1f 24.) This was so despite
'£ !| the fact that IBM's Product Test organization was of the view that the
'" System/360 Model 30 central processing unit was farther behind in their_
testing procedure than any of the other System/360 central processing
units announced in April 1964. (PX 1107, p. 7.)
.7
S
:j ** By the time 360 was announced, engineering models of all the
: q ; | processors had been built (Brooks, Tr. 22695-96); full instruction set
k ii compatibility across the five processors had been achieved (Brooks,
?rr j.Tr. 22785) ; a complete processor had been built usinq SLT tech-
" M -inology and demonstrated to establish the feasibilitv of the
. 7 ;jnew circuitry (Hughes, Tr. 33952-55; JX 38, 1f 7, p. 5; DX 4800);
~ 'jmany thousands of SLT modules had already been produced (DX 4796, p. 8)
~ 1 most of the processors and some of the peripheral equipment were in
— jthe early stage of product test (McCarter, Tr. 88383; JX 38, If 19);
,- ilall, or almost all, the memories had undergone technical evaluation
^ [testing (Brooks, Tr. 22699) ; microprogramming and multiprogramming had
m4 Ibeen tested on the Model 40 (McCarter, Tr. 88382-83); and four esti-
-" [mating, forecasting and pricing cycles had been completed (DX 1172, p.
;2). Product Test had been involved with the development from the
- 5 :j beginning (McCarter, Tr. 88375; DX 1165): by the time of announcement,
I
:[
'•! -353-
L
2
5
4.
3"
7
3
a
La ;
Li
12
13.
14-
L5
IS
17
IS
19
2C
In September of 1963, Evans wrote to DSD President Kennard:
"NPL is good — it is simple and powerful — it is ready enough
— proven enough. IBM should go forward with . . . full
announcement in the first or second quarter of 1964 with .
programming systems committed." (DX 2983.)
Although the SPREAD Report had not recommended announcing the entire
NPL family at once, by December 1963 it was plain that there were
powerful reasons for doing so. On December 27, 1963, Evans proposed
that the NPL family be announced as a group in March 1964:
"[Tlhe customers must better understand the abilities
of the architecture and conversions necessary. It would
be unwise of us to announce systems sporadically in an
effort to optimize market penetration or profit. It is
proper that IBM announce all the systems in a group so
that our customers have the benefit of the family and
can properly plan." (DX 4815; see also Evans, Tr.
101072-75.)
Less than one month later, Brooks wrote to Gibson, Haanstra and
Kennard, stating that the equipment was "technically ready for
announcement" and recommending announcement on April 7. (DX 1172.)
He emphasized that System/360 " must be announced at one time" (id.,
p. 3) :
"Piecemeal announcement would utterly confuse and misguide
the customer in his planning. He could not make the best
selection from the available models until all the models
are announced." ( Id. ; see also Knaplund, Tr. 90486-88;
Brooks, Tr. 22782-84.)
~ i! thousands of tests had been made and "literally hundreds of problems
^ ;!and potential problems" had been identified and resolved. The compon-
^— j entry, systems and product testing program already completed was more
„_ i| extensive than the entire program IBM had previously undertaken for
^ {any system. (McCarter, Tr. 88390-93; Evans, Tr. 101065-66, 101082;
|DX 1172, pp. 2, 5; see also DX 4815.)
| RCA, Honeywell and GE all announced systems that were, by com-
- 5 Jparison, in an embryonic stage of development. (See Spangle, Tr.
14997-99; Weil, Tr. 7232-35; McCollister, Tr. 9635-41.)
1
1
■I -359-
Knaplund testified that he "understood that simultaneous announce-
ment . . . would place an unprecedented load on the development of
manufacturing resources of the product divisions " • However, the
•advantages outweighed the risks* Since IBM was unquestionably
"going to produce a compatible line, only by fully informing
customers as to the full compatible range, including prices and
functional specifications, could they evaluate properly IBM's
offering. "It was my business judgment that partial announcement
by IBM would result in customer confusion, superseding orders
following subsequent IBM announcements , and churning of the order
backlog in IBM's production schedules." (Knaplund, Tr. 90486-88.)
The March or April announcement dates recommended by
,4
5
5
7 ;
,3 ;
a
m ! l Evans and Brooks were virtually mandated by the first shipment
j| dates planned for the 360 processors, which ranged from June
1965 for the 2030 to January 1966 for the 2070. (See JX 38,
1 24.) It was "generally industry practice on most computer systems
[at that time to announce a system at least a year, and frequently as
:jmuch as two years, ahead of the actual first delivery".* (Weil, Tr.
II
;(7064; see McCollister, Tr. 9635, 9641, 9646; Hangen, Tr. 10761-62;
3 i|
;JKnaplund, Tr. 90483-84; PX 355, pp. 33-36; PX 2226A,pp. 13, 19, 27;
JPX 2432, pp. 19, 22, 28; DX 573; DX 4769; DX 4774; DX 8962.)
1 " * There were "practical reasons" for this procedure from both the
23 'manufacturer's and the customer's viewpoint, each of whom needed time
(to prepare for delivery and installation. (Weil, Tr. 7064-65;
?A ;|Withington, Tr. 58738-46; J. Jones, Tr. 79034-36; Akers, Tr. 96537-40;
•DX 3726.)
^ 1
t
i
I • -360-
i
i '
i
!
■\
•l
1* ' Such lead time was particularly important in the case of
2. || System/360. Thus, Brooks wrote in January 1964: "The breadth of
System/360 and the number of innovations, particularly in gross
systems concept, will require substantial lead time between announce-
ment and proper installation." (DX 1172, p. 1; see also DX 3726;
DX 4815.) That time would be necessary to:
(1) permit customers to replan their applications and take
advantage of 360* s new concepts such as file orientation,
communications facilities and large memories ;
(2) permit customers to assimilate the "sheer amount- of new
abilities, new options, new specifications, and new prices"
that 360 would provide and select the best configuration of
equipment to perform their applications;
(3) permit IBM and customers to educate their personnel and
prepare them for proper installation and maintenance
of 360;
(4) permit IBM to avoid deferred installations and conse-
quential inventory build-ups;
;j (5) permit customers to determine the need for and submit
19 :i-
;[
;i RPQs for special requirements; and
20 J
:i (6) permit customers to prepare their physical sites for 360
22. -j -
j installation.
(Withington, Tr. 58738-46; J. Jones, Tr. 79034-36; Knaplund, Tr. 90483-
88; Akers, Tr. 96537-41; DX 1172, pp. 1-2; DX 3726; DX 4815.) As
24.
'Southern Railway's President was advised by his EDP staff in 1964,
25 j
1 -361-
3:
5
Si
7 i
3
a
10
11
12. !
13-
!A !.
15
lo
17
IS
r ;
m
1 |
i
I
t
J I
z
7 j
3 j
i
a |
I
a I
i
i \
2 •
^»
J
,6
.5
"there is always a year to 18 month delivery' lag from ordering to
delivery. This amount of time is usually . . . required for planning
and programming" . (DX 3726.)
Over the course of the NPL development, there were numerous
proposed announcement dates considered by various IBM personnel,
ranging from mid-1963 to mid-1965. (Brooks, Tr. 22796; JX 38, 1( 15,
p. 8; PX 1079? PX 1092; DX 1404A, pp. 57, 70, 119 (App. A to JX 38);
DX 4782; DX 4786; DX 4790; DX 4814; DX 4815.) In December 1963, Paul
Knaplund was assigned responsibility for assembling the technical
evaluations, forecasts, cost analyses and profit projections that IBM
top management would need to address the 360 announcement decision.
Beginning in January 1964, he conducted weekly meetings with IBM line
S and staff management to identify and assess the magnitude of outstand-
ing problems and outline programs to solve those problems , so that he
and they would be prepared to make judgments and advise top management
on the advisability of proceeding with the 360 announcement. (Knaplund,
i|Tr. 90474-77.)
On March 18, 1964, IBM Chairman T. J. Watson, Jr. made the
!j final decision to announce all of the models of the new line simul-
:ltaneously on April 7, 1964. IBM's Product Test Department did not
isupport the April 7 announcement — all other departments whose effort
II '
'was required to provide the products, features and services offered in
?*?
~ ; jthe System/360 announcement did support it. (Gibson, Tr. 22648;
"* JBrookG, Tr. 22799-800; Hughes, Tr. 34003; Knaplund, Tr. 90483, 90493;
.1
Zt jE. Bloch, Tr. 93311; JX 38, If If 18, 22; DX 1165; DX 9161.)
■I
I -362-
•i
i
i
z
3
4-
f
5
7
3
9
ia
n
12
13
14-
15
Evans, Brooks and McCarter explained the organization and
role of the Product Test Department, in addition to all the other
testing that was done at the time within IBM; how IBM management
used the Product Test position to isolate problems and challenge
the product development personnel to determine how they would solve
those problems; and how Product Test, after it took its non-support
position, later supported the shipment of System/360 to IBM's
customers. (McCarter, Tr. 22568-70, 88362-93, 88434-55; Brooks, Tr.
22786-88, 22850-53; Evans, Tr. 101065-66, 101083-95, 101174-78;
PX 2126, pp. 2-5, 35-37; PX 4005; DX 1165; DX 1172, pp. 2, 5; DX 1409;
DX 4815; DX 8083.) As G. B. McCarter* testified,
"It did not follow from Product Test's non-support of
March 16, 1964, that IBM could not or would not deliver what
it committed to customers. ... To the contrary, Product
Test's input was one of the mechanisms, like internal targets,
designed to ensure that it would." (Tr. 88404.)**
In fact, the processors announced on April 7, 1964, were
-- ji all shipped on or before the dates estimated for shipment at
announcement, except that the 2060 and 2062 on the one hand, and
the 2070 on the other, were superseded by faster memory versions
17 |
IS \
I
19 !
22 ,| * McCarter was DSD Manager of Product Test, and was the person who |
j presented the position of the Product Test organization for all IBM • J
21 ;j divisions to IBM management prior to System/360 announcement. (McCarter!
•i Tr. 88373, 88380-81.) , j
22;! j
\ ** Prior to 360, there had been numerous occasions on which IBM j
23 j announced products without Product Test support, including the 140 3 ;
,! printer; 1302 disk file; the 709, 7090 and 7074 systems; and more than
24- ! two dozen software programs. (McCarter, Tr. 38371-72, 88602-05; Evans, j
I Tr. 101093-94; DX 4768; DX 7630; DX 9005.) j
-363-
1 called the 2065 and 2075, respectively, which were delivered on or
2 | before the dates planned for their predecessor processors in April
3 ; 1964. (JX 38, If 24.) Those first shipped systems, as planned, were
4 • made available with the simpler operating systems offered with 360.
5 ; (Brooks, Tr. 22853.) However, as we mentioned earlier, there were
6 '• "significant schedule slippages in OS/360 software", (the most ad-
7 vanced operating system for 360) which meant that some customers
8 "received the full announced capabilities later than originally
9 '. planned". (JX 38, II 25; DX 4740: Evans, Tr. (Telex) 3933-34; Welke,
10 I Tr. 19410, 19631; see also Enfield, Tr. 20947-43; PX 4834, p. 23.)
11 The problems with OS/360 occurred even though Product Test "cumula-
12 tively did more testing of OS/360 than we ever had before for any
13 set of programs for a particular system" (McCarter, Tr. 88390-93) ,
14 and despite the fact that IBM's programmers believed prior to April
15 7, 1964 that they could produce OS/360 "in the way that it was ori-
16 ginally intended". (McCarter, Tr. 88390-93; Evans, Tr. 101119; DX
17 | 5609.) IBM, like the rest of the industry,* misjudaed the "enormous
!
18 i complexity" of developing complex operating systems. (Perlis, Tr.
19 ]320, 2001-03; Spangle, Tr. 4997-99; Weil, Tr. 7215-21; McCollister,
20 9696-98; Welke, Tr. 19281-82; Brooks, Tr. 22762-63; Withington, Tr.
21 55914, 56729-30; McCarter, Tr. 88390-92; Evans, Tr. 101119.)
22
23 *Withington testified that "all manufacturers attempting the most
advanced systems programs in that time had difficulties". (TR 56729-
24 30.) For example:
25 (a) Burroughs announced but never delivered the 38500 and
B7500, in part because of software problems. (Perlis,
-364-
tj
2 I
i
!
5 !
I
I
S ;
In a way, the modularity and standard interface of the
System/360, which made hardware testing easier, as already discussed
2
W
II i|
i.
13
16 ,
17 [
Tr. 1320-21, 2001-03; Withington, Tr. 56599-600.)
(b) Univac was compelled to delay the introduction of
its EXEC 8 operating system for two to three
years. (Perils, Tr. 2001-03.) Earlier, Lawrence
7 jl Livermore Laboratory was compelled to rewrite "com-
| pletely" the software that Univac had provided with
3 i the LARC computer because the Laboratory was "not
satisfied with it". (Fernbach, Tr. 517-18.)
(c) Xerox "had difficulty producing the UTS [operating]
system that [it] had announced". UTS was delayed
for several years, costing XDS several millions of
dollars in revenue. (Perils, Tr. 2001-03; Currie, Tr
15303, 15352-54.) XDS also experienced delays
in its XOS operating system. (Currie, Tr. 15704.)
(d) The MULTICS operating system was never delivered
by GE, even though GE, MIT and Bell Labs believed
it could be feasibly designed. Honeywell finally
completed the development three years behind the
original schedule. (Weil, Tr. 7232-35; Wright, Tr.
13373-76; Withington, Tr. 56730-31.) GE also
had difficulty in making GECOS perform to their
customers' satisfaction. Three different versions
were eventually constructed, and none ever met the
advertised capabilities. Because of those diffi-
culties GE withdrew its Models 625 and 635 from
IS :[ the market for a year or two in late 1966 or early
:| 1967. (Weil, Tr. 7215-21; Withington, Tr.
15 :j 56730-31.)
22 .; (e) The Honeywell 8200 was unsuccessful, in part,
.1 because of software development difficulties.
21 ;| Honeywell had to spend "large amounts of money,
i more than we had planned" to develop the soft-
22.! ware. (Spangle, Tr. 4997-99.) 'Honeywell also
;j took longer than anticipated to develop its
23 j Series 60 line because of "difficulties in
I developing software and microprogramming".
24 1 (Spangle, Tr. 5008.)
25 (f) RCA's TSOS was delayed "on the order of six to
-365-
(see above, pp. 360-62) , made software testing harder. It allowed
customers great flexibility in the range of configurations which they
could choose, and that, coupled with the wide variety of ways in which
OS/360 could be used, led to "a very complex hardware-software system"
which was literally impossible to test adequately. (McCarter, Tr.
88544-45.) As Enfield testified:
"Systems software by its nature cannot be adequately tested
in a single environment but must in fact be tested . . .
in a user environment in order to establish the many
different types of configurations, the many different types
of generation, options, the many different types of operat-
ing environments.
"If you were to take the various permutations of the
options available to the user, the number of different
tests that would have to be performed [in testing systems
software] would exceed the time available for testing.
I am talking about millions of different permutations and
combinations of features that can be selected by the users .
To test in each of those environments would preclude the
issuance of first release of any operating system ....
[bjecause as soon as you got around to testing the 999,000
somebody would come out with another option and you'd have
to go all the way through it again." (Tr. 20294-97; see also
Perlis, Tr. 1347-48.)
Only by expending "considerable internal efforts" was IEM
j
VI "i
:j twelve months, possibly more", and performed poorly
£ j and unreliably. (McCollister, Tr. 9694-95, 9707-08;
| Rooney, Tr. 12132-34.) RCA's VMOS also experienced
jt .; delays, which were estimated to constitute a "poten-
tial problem" of some $2 million in monthly rentals from
lost accounts plus a loss from delayed installations of
~ j $3 1/2 million. (Rooney, Tr. 12335-36, 12349-50, 12358;
T2 ! Conrad, Tr. 14088-89, 14133; DX 872, p. A.) The
j difficulties with and instability of TSOS/VMOS
7;t ; | "endangered [RCA's] position with any -customer who
j had equipment on order and who planned to use this
5S ! operating svstem" . (McCollister, Tr. 9704-05,
~ -! 9710-11.)
-366-
L
2:1
able to remedy the problems with OS/360 — but IBM did so and provided
customers a "very sophisticated, very complex software system, a
software system that permitted the customer a great deal of flex-
* | ibility . . . the customer could do a great deal with a minimum amount
-'{of effort", which in turn caused System/360 "to show steadily in-
creasing performance relative to competition and remain saleable
longer". (Perlis, Tr. 1887-88; Palevsky, Tr. 3180; Rooney, Tr. 12576;
Currie, Tr» 15186; Welke, Tr. 17308-13; McCarter, Tr. 88389; PX 1900,
pp. 3-4, 8; PX 4833, p. 16; PX 4834, p. 23.)
b. 360 's Success and Impact on IBM. System/360 was
launched on April 7, 1964, and the internal doubts about its reception
were soon dispelled* (See Knaplund, Tr. 90515; DX- 4740: Evans, Tr. (Telex)
3932-33.) Orders for the systems "far exceeded IBM's forecasts"
(Gibson, Tr. 22636-37; Case, Tr. 73258; Knaplund, Tr. 90547; Evans,
Tr. 101123; Cary, Tr. 101780-81; JX 38, 1f 28; PX 1900, pp. 7, 10;
DX 9331) and exceeded by thousands IBM's production plans which were
Ibased on those forecasts :
20 \
Gross Orders
Estimated Actual Booked
~ |l965 589 668 4,487
1966 2,897 3,132 • 4,526
22 1
** 11965 & 1966 3,486 3,800 9,013
_, ii (combined)
24. i
25 i
-367-
(JX 38, <f 28.) By October 1966, IBM's 360 order backlog represented
an income of "almost three times . . . [IBM's then-current,] worldwide,
annual sales of all products". (px 1900, p. 10.)
As we discussed earlier, IBM management authorized sub-
stantial increases in plant capacity prior to 360' s announcement in
order to meet anticipated production and delivery requirements —
including the establishment of an SLT manufacturing plant in East
Fishkill, N.Y., and the addition of a new building at IBM's Endicott,
N.Y., plant site for the manufacture of SLT cards and boards. (See
above, pp. 344-45.) It was management's judgment that these manufac-
turing capacity increases "adequately provided for the component
and box production volumes required to support the System/360
announcement together with planned future announcements". (Gibson, Tr.
4JJ 22635-37; Knaplund, Tr. 90545-46; E. Bloch, Tr. 91895-96; DX 7691,
-! p. 4; DX 9333.) However, because the total orders were far beyond
~ \\ what was forecast and because larger size processors and more memory
1
a
and peripherals than anticipated were being ordered, the demand for
SLT modules also far exceeded IBM's expectations. (Knaplund, Tr.
J
,S :i
,_ 1 90547; E. Bloch, Tr. 91899-906; Dunlop, Tr. 94774-75; DX 9331; DX j
J :j !
;l 9332; DX 9333; DX 9334.) By May 1964, only a little more than one \
:Q -1 !
. i month after announcement, the projected "maximum annual ^fcdule Be- j
d \ I
■j quirements" had increased from 70-90 million to 130-190 million. (DX j
•I 9331; see also E. Bloch, Tr. 91899-900, 91905-06; Dunlop, Tr. 94774-
22 :i i
i 75; DX 9332; DX 9333.) i
2-i . !
• It was plain that the manufacturing capacity planned at j
'j ;
i
■i ;
1
; | -368-
L
2
2
3
s
71
3
9
10 !
li
121
13-
14
announcement would be insufficient and IBM began moving to meet the
increased demand. By the third quarter of 1964 additional component
production capacity was approved as an addition to IBM's Burlington,
Vermont, plant site, and plans were initiated for additional assembly
plant locations. By the end of 1964, IBM top management had approved
expansion of the Federal Systems Division's Owego, N.Y., plant "to
increase manufacturing capacity for SLT cards and boards"; and in
the first part of 1965, two new plant sites in Boulder, Colorado,
and Raleigh, North Carolina, were approved "to increase IBM's
overall EDP manufacturing capacity". (Knaplund, Tr. 90547-48; E.
Bloch, Tr. 91905-08; Dunlop, Tr. 93670; PX 5771, p. 28; DX 9038.)
In addition, IBM provided special tools and training to Texas Instru-
ments employees so that Texas Instruments might serve as an additonal
source for SLT components. (E. Bloch, Tr. 91908.)
By October 1965, IBM announced that it was "completing more
than three million square feet of new manufacturing space" to meet
requirements for System/360 — including plants in Boulder, Colorado;
Raleigh, North Carolina; Montpellier, France; Vimercate, Italy; and
expansions of existing facilities in Owego, Fishkill and Endicott, New
iJYork; Burlington, Vermont; and San Jose, California. (DX 9038.) New
20 1
^plants were later added in Boca Raton, Florida and Brooklyn, New York.
16
17
is
21 j|
j (Dunlop, Tr. 93670.)
IBM also began hiring substantial numbers of new employees
(Between year-end 1964 and year-end 196 7 IBM increased its work
m * \
jforce by approximately 50% — adding more than 70,000 new employees.
25 ; ■
! -369-
(PX 5771, p. 3.; DX 13680, pp. 3-4; see also Knaplund, Tr. 90549-50;
2 ! Dunlop, Tr. 93670; DX 4740: Evans, Tr. (Telex) 3934.) Evans testi-
3 : fied that it was "an enormous job" to get the supply of parts flowing
4 \ hire the people and train them in order to meet 360 commitments.
5 ; At one point, IBM "even rented a circus tent to temporarily store
6 • parts" until more permanent facilities could be secured. (Knaplund,
7 Tr. 90549-50; DX 4740: Evans, Tr. (Telex) 3934.)
8 In January 1965, IBM combined all product division manu-
9 facturing functions in a single manufacturing division. It was
10 ! believed that "by unifying responsibility for scheduling and produc-
11 ing all the principal System 360 equipment, . . . manufacturing effi-
12 ciency could be increased and information flow accelerated".
13 (Knaplund, Tr. 90548-49.) The Systems Manufacturing Division (SMD)
14 was thus created, with former GPD President C. E. Frizzell at its
15 head. ( Id. ) By June 1965 Frizzell reported to IBM management that
16 the production buildup would enable IBM to meet product shipments
17 ! committed to customers. (Knaplund, Tr. 90550-51; DX 1154; DX 1155;
18 J see also E. Bloch, Tr. 91915; DX 9333.)
19 Within a few months, however, an "unforeseen" technical
20 difficulty developed in the production of SLT technology. (Knaplund,
21 Tr. 90551-52; E. Block, Tr. 91915-18.) The problem took about three
22 months to sold, despite intensive efforts by IBM to do so, and the
23 S delay put I3M several months behind the schedule for SLT production
j
24 j needed to satisfy existing customer commitments. (Knaplund, Tr.
i
25 I 90551-52; E. Block, Tr. 91917-19.) This was reported to I3M Chairman
-370-
L
2.
3
4*
5
7
3
9
10
II
12
13
14- I
Watson , who immediately informed IBM's Board of Directors and issued
a public statement advising that "during 1966 most System/360' s
will be delivered 60 to 120 days later than originally scheduled" .
(Knaplund, Tr» 90551-52; DX 9038.) Knaplund testified that, but
for the unanticipated production problems, System/360 shipments at
that point "would have continued on the committed plan". (Tr.
90552-53.)* In the end, although many 360 hardware deliveries were
made as scheduled and committed, there were some significant schedule
slippages despite all of IBM's efforts to prevent them. (See
Knaplund, Tr. 90849-54; JX 38, U 25.)
The production, delivery and installation of System/360
required a massive effort on IBM's part, which placed a severe
strain on the corporation. (Cary, Tr. 101359-60; PX 1900, pp. 4, 8;
DX 4740: Evans, Tr. (Telex) 3932-34; DX 8386, pp. 107-08, 111; DX 13677,
p. 5; DX 13678, pp. 6-7.) In November 1965, Watson wrote to all
I3M managers: "We're carrying out an assignment that in many
„_ ! [ respects is one of the largest and most complex ever given to an
;i
I industrial electronics organization — almost a complete replacement
of our principal product line". (DX 8886, p. 107.) It was a task
IS
18 i
13 i
„ f that some in IBM likened to "trying to swallow an elephant".
20 !
m I (Carv, Tr. 101359.)
J As we have discussed, IBM had to build new facilities and
22 .
23 I
| * Despite the problem, IBM's SLT output for 1965 was higher than
2- i that planned in April 1964; IBM was also able to achieve a 74%
| increase of production in 1966 over 1965. (Knaplund, Tr. 90943-46;
25 J E. Bloch, Tr. 91917.) In May 1966, the 2,000th System/360 was
! shipped. (JX 38, If 27, p. 10.)
1
'I -371-
r
i
r
3 ;
i
?!
" i
a j
i
3 |
i
4>;
hire and train many new employees. The size of the job was com-
pounded by the software difficulties with OS/360 ► IBM placed a
"top priority" on the solution of those problems and, at its peak,
M' had over 1000 people working on. OS/360 ~ Some 5000 man-years went
into its design j construction and documentation between 1963 and
1966. (PX 468, p. 31? DX 13677, p. 7; see also DX 4740 : Evans, Tr.
(Telex) 3932-34.)
The breadth and complexity of System/360 together with
the new, advanced applications for which it could be used required
IBM to provide "the most extensive total programming systems support
ever developed". (DX 13677, p. 7.) It also meant that IBM would
have to provide customers more assistance than ever in installing,
understanding and applying 360 and all its revolutionary new concepts.
(Case, Tr. 73590; Evans, Tr. 101127-28; DX 1172.)
The need to expand quickly to meet the unforeseen explo-
s : sion in demand for 360, to hire and train new employees and to
it support customers' in their installation and use of the new systems
'\ placed "tremendous capital demands" on IBM. (Cary, Tr. 101525-26;
>3
it
_ 1 DX 8886, p. Ill; DX 13677, p. 5; DX 13678/ P- 7-) During 1964 IBM
:| had prepaid $160 million in debentures and promissory notes. (PX
•!
„ i 5771, p. 36) As a result, it did not have sufficient money on hand
/ 2. i!
'■i to finance the required expansion and had to raise it. In 196 6 IBM
:| raised approximately $371 million through an eauitv offering, the
1 first such offering since 1957. (DX 13339, p. 20; DX 13678, p. 39.;
:J IBM Chairman T. J. Watson, Jr., explained to IBM's stockholders:
-372-
L
Z
z
4-!
5
7
3
ii
12
13-
14
15
IS
17
is!
21 !
"Because of the plant construction program and the
System/360 production build-up, 1966 required a
worldwide investment, of approximately $1.6 billion in
rental machines and parts, factory, laboratory and
office equipment, and land and buildings. To help
finance this expansion in our business, additional
capital stock was offered to stock holders last June.
$371 million of new capital was raised in this
manner."* (DX 1367 8, p. 7.)
In 1966 and 1967 IBM raised its lease prices and decreased purchase
prices by 3 percent. (PX 4481A, p. 1.) A "major consideration"
for the change was "to encourage purchase, and thus, increase the
amount of cash needed to finance higher-than-anticipated demands
for the 360". < PX 6153, p. 2; Cary, Tr. 10X525-26.)
IBM's multi-billion dollar investment yielded fantastic
rewards , changing the face of IBM and of the computer industry for
all time. Chairman T. J. Watson, Jr. called 360, at the time of
announcement, "the most important product announcement in company
history". (PX 1900, pp. 7-8.) He could not have been more right.
System/360 was a "phenomenal success", perhaps the greatest "in the
history of American industry". (Cary, Tr. 101781.) As we have
already seen and as IBM's current Chairman, Frank T. Cary, testified,
"customers loved it", and "ordered it in quantities way beyond
In 1965, IBM had reported that "the plant expansion program
i and System/360 production required a record worldwide investment of !
22 '.» $1.1 billion in 1965 for rental machines and parts, factory and i
office equipment, and land and buildings' 1 . (DX 13677, pp. 6-7.) [
25
-373-
I
3
7
3
3
I 1
anything that we had anticipated". (Tr. 101781.) 360 shipments
exceeded by more than double the estimates made prior to announce-
ment. (Case, Tr. 73258.)
The effect on IBM was profound. At year-end 1963, when
the production buildup for 360 was begun, IBM employed 138,281
people worldwide (PX 5771, p. 3); by year-end 1969, IBM's employ-
ment had nearly doubled — to 258,662. (DX 3364, pp. 3-4.) Over
that same period, IBM's manufacturing floor space in the United
States climbed from just over six million square feet to more
than fourteen million square feet — more than double. (DX 1396 3
DX 13964 , pp. 1-3.) At year-end 1965, before volume shipment of
360 had begun, IBM had worldwide revenues of $3,572,824,719 (DX
13677, p. 5); by year-end 1970, IBM's worldwide revenues had
increased more than two times, to $7,503,959,690. (PX 5767,
p. 3.) Just prior to the 360 announcement, IBM had approximately
11,000 systems installed in the United States. By the time 370
was announced, that number had tripled to approximately 35,000.
In the interim IBM's corporate growth, revenue and profits were
"way beyond anything that [IBM] had anticipated". (Cary, Tr. 101360,
101781; DX 4740, Evans, Tr. (Telex) 3934-35.)
These numbers demonstrate the extent to which IBM's
jj success, as it stood on the threshhold of the 1970s, was the
i
I result of an overwhelming acceptance by users of System/3 60 and
I
-374-
Z
z
4»| "Observers have characterized the 360 decision as perhaps
the biggest,, in its impact on a company, ever made in
American industry — far bigger even than Boeing ' s decision to
go into jets, bigger than Ford's decision to build several
million Mustangs.
3
e
7
a
la
u
IZ
14
15
Iff!
17 ';
» I
20 1
21 ij
1
22 ;!
of IBM's ability to put the system into production and install it
in unprecedented and unforeseen numbers. As T. V. Learson wrote
in October 1966:
n i
2±
25
"IBM. has certainly not been the same since, and never
will be again". (PX 1900, pp. 8-9.)
-375-
36. Initial Competitive Responses ■ to System/360 . The
System/360 announcement and its subsequent success provoked a host
of competitive responses from a variety of different sources, including
systems suppliers, leasing companies, peripherals manufacturers and
[software houses. As we discuss below, System/360 spurred the rapid
growth of leasing companies, software suppliers and peripherals
suppliers in particular, and each applied increasing pressure on IBM
as they grew in strength. In this section, we review the more
immediate actions taken by a number of systems suppliers. (These
actions are discussed in more detail below, pp. 377-84.)
We saw above how, by the time of the System/360 announcement,
IBM's earlier computer lines had been "leapfrogged" by competition,
L
I
I
3
9
a
x
•7 |
"" I
-» | and how System/360 gave IBM a price/performance advantage over
competitive machines. Indeed, as Knaplund testified, it was understood
that the price/performance advantage of System/360 as measured by
! IBM employees understated the true superiority of System/360 compared
to competitive offerings.* (Tr. 90503-05.) The System/360 announce-
ment, therefore, forced IBM's competitors to reduce prices or
j increase performance in order to remain competitive. Weil of GE
i
said in June 1964:
J I
,3
3
| "The entire competitive picture in the information
II ; | processing business at this time in 1964 is characterized
\
2 :!
£3 ;• * According to Knaplund, the methods available within IBM at the
time for making price/performance comparisons could not adequately
i- (evaluate several advantages of System/360: the use of disks, the
! improved reliability, the factor of compatibility and the software
If I support. (Tr. 90504-05, see Tr. 90506-09.)
.i
I -376-
Li
2 j
z\
I
3 I
|
7 I
3 I
9 I
La-;
II
IZ
n.
LO-
IS
by the impact of the IBM System/360 . . . announcement and
by the reaction to this announcement of our competitors.
"The System/ 3 60 is an excellent product line with out-
standing peripheral offerings." (PX 320, pp. 12-13.)
The result, according to Weil, was that it was "no longer possible
to offer equipment with a significant advantage over IBM". ( Id. ,
p. 14.) In July 1964, Learson reviewed the price reductions in the
industry that had taken place since the System/ 360 introduction and
wrote:
"There, can be only one conclusion; namely, the cost/
performance of computers today is less than it has been
and . . . the price structure surrounding the main body
of our line is threatened by: (a) Present day cost[,]
(b) New technologies, as typified by NPL[.] Perhaps
what we are missing is that NPL was a price reduction of
30-50%, so that competition is forced to come along with
us." (DX 1525.)
And they did, with price reductions, product announcements or both.
In order to be competitive, most companies tried to price their
la ;l products to achieve anywhere from a 5 to as high as a 40 percent
17 ij price/performance advantage over IBM's 360 line. Despite the
IS ;! acknowledged difficulties of comparing the performance of systems
19 1 (see, e.g., Palevsky, Tr. 3269-71; McDonald, Tr. 4207), such a
20 ;l pricing policy was common among competitors , who felt they needed to
21 -j offer something better than IBM to attract customers. (See below,
■i .
22,'tpp. 377-84) IBM monitored these reactions in some detail, and undertook
1
22 • to respond,
i
2^ 'j a. RCA . RCA both reduced prices on its current products
2f [and shaped its planned new announcements in reaction to 360.
\
} -377-
j
i
■i
L
I
I
M
2
T\
3
9 !
a
i \
2 i
i
f:
17 ;:
TQ
20
In approximately May 1964, according to internal IBM reports, RCA
reduced the price of its 3301 between 20 and 35 percent. (PX 2956,
p. 1; DX 1525; see also PX 4829, p. 19.) Within IBM the price
reductions were seen as "drastic", as "the first significant com—
«
petitive reaction to System/360" and as making "the 3301 very
competitive in the model 40/50 area". (PX 2956, p. 1*)* Withington
wrote that the "primary reason for the price reduction . . . would
seem to be a requirement for a competitive product during the interim
until RCA announces its 'counter- 3 6' * efforts". (PX 4829, p. 19.)
Soon after, RCA announced the Spectra 70 Series, which was
designed to be compatible with the 360 line. (See below, pp. 551-58.)
The preliminary design of that series had started in 1963, with
"[mjajor design efforts. . . . under way by the latter half of '64".
(Beard, Tr. 8459; see p. 551 below.) The strategy of compatibility
with IBM equipment had been considered prior to the 360 announcement
(Beard, Tr. 9113-14), and was firmly decided "within two weeks,
three weeks at the most, after the announcement". (McCollister, Tr.
9630.) By making its Spectra 70 compatible with IBM's System/360,
RCA hoped to be able to persuade 360 users to move to Spectra: it
was "aimed primarily at the IBM 360 series range of computers".
(Beard, Tr. 8459; see pp. 552-58 below for a fuller explanation of
21 |
j this strategy.)
..1
I
I * In July 1964, Learson interpreted the 3301 price decrease as
\X something forced upon RCA by the "bad price/performance ratio" of the
3301 product and its failure to sell. (DX 1525.)
-378-
3
10
II
12
13
L£:
15
is i
17 1
IS
L ; Within IBM the announcement of Spectra was noted in a
Z l memorandum from C. E. Frizzell, President of GPD, to T. J. Watson,
3 I Jr. Frizzell wrote that the series offered better price/performance
4* |[ than IBM in CPU -memory speed, magnetic tapes and high speed printing
y- |t but assured Watson that he was ,r moving rapidly to meet this challenge
3 : and expect to respond effectively in the very near future" . (DX
7 If 960.)
b. GE. Weil testified that at the time of the 360
announcement, GE assessed 360 as a "very strong and very competent"
competitor to its current system, the GE 400. (Tr. 7060.) As
noted earlier, his own assessment in 1964 was that it was "no longer
possible to offer equipment with a significant advantage over IBM".
(PX 320, p. 14.) Then he reported at the same time (June 1964) that
GE was planning to announce a new series of magnetic tape units
"which will permit adjustment of our 400 line system prices to
increase our competitiveness". ( Id. , p. 16.) IBM sources reported
that GE did reduce prices on the 400 in reaction to System/360.
Lear son wrote in July 1964:
• "GE has not officially reduced prices, but they
IS j|. are selling their 400 line at 18% off. They have also
reduced their extra shift to a 10% charge.
2°
:i "Further, GE is selling their 635, a competitor to
21 •[ . the 7094, at no extra shift charge." (DX 1525.)
22 !| A September 1964 Competitive News Release from the Data
»i
23 | Processing Division's Commercial Analysis ' Department confirmed
24- } price reductions of 8%-15% and went on to say, "The price
25. '
.i
•I -379-
I i
z
3
a i
6
7
3
9
ia
ii
12
13
14
15
is ;
17 i
!
18 ■
i
19 ;
20 ;
21 I
22
24
25
reduction gives the GE 400 a price/performance advantage over com-
parable System/360 configurations." (PX 2966, p. 3*) However,
Knaplund felt, that the price reductions were necessary for GE to
remain competitive after System/360. ( Id. , p. Rl; see below, pp.
490-93.) A subsequent price/performance evaluation made within IBM
concluded: "While the recent price reductions have improved GE's
position, the System/360 Model 30 retains its price/performance
superiority." (DX 13445.)
General Electric announced its 600 series in the summer of
1964. Although planned long before the 360 announcement to displace
IBM's 7090 and 7094 computer systems (see below, pp. 493-505), GE
called the 600s a "family . . . for business, scientific and real-
time use". (DX 491, p. 1.) Weil had compared the 600 series against
the 360 line in a June 23, 1964 internal GE presentation and con-
cluded that the 600 is "either just a little more favorable or just
a little less favorable than comparable members of the 360 series.
We are, however, able to deliver our equipment a year earlier than
IBM". (PX 320, p. 16; see below, pp. 493-505.)
!j GE saw itself as being able to capitalize on one of the
j risks IBM had taken with the 360 — the risk involved in making the
:l older lines obsolete. Weil testified that the computer group at GE
jl was "initially at least overjoyed with what had occurred because it
meant right at the time we were introducing a system designed to
displace 7090s and 7094s, IBM had itself abandoned 'che 7094 and 7090
computer series and brought out an entirely different computer series,
-380-
L
2
3
4.
and it was our belief at that time that it would be easier, if you
were a user, to convert from the 7090/7094 to the 600 series than it
would be to convert to IBM's new 360 series. We regarded that as a
fortuitous occurrence and potentially to our advantage." (Tr. 7060-61.)
The user of the 7094 was "forced ... to either go to a 360 or to
some other competitive system, and we were sitting there with a system
designed to make that conversion as easy as possible." (Tr. 7062.)
c. CDC . According to Weil, CDC also reduced prices in
response to System/360. (PX 320, p. 16.) At IBM, Learson analyzed
CDC ' s behavior as follows :
7
3
9
ia
Ll
12
13
14- ; And Withington wrote
"CDC followed [360' s pricing] with a price reduction
of their 3600, which was no longer competitive with the
360-Model 70. In dropping the price of the 3600, they
had to keep their deck of cards in order and so moved the
3200 and 3400 downward. Reductions of 20-40% were made."
(DX 1525.)
"Control Data's main reliance is on price; apparently
its intention is to provide a lower cost answer to every
IS '! System 360 model. After the System 360 announcement, the
;i price of every existing Control Data computer was reduced,
U ;| and the prices of the later models are still lower. . . .
This should unquestionably help Control Data's position
^g J because . . . the market is becoming increasingly price-
conscious." (PX 48 29, p. 21.)
:!
:|
19 I
Several months later, CDC announced new members of its
•] current product lines — the 6000 and 3000 series. The formal
21 i
j announcement of the 6400 (a "scaled down" 6600) and the (never
22 ; S
j delivered) 6800, to go with the existing 6600, was made in mid-
23 -i
(December 1964. (Norris, Tr. 5626, 5965-67; DX 319, p. 1.) The 3300
~" jand 3500 were announced in 1965. (PX 355, p. 35.)
I
I.
i
! -381-
z
3
4.
3 !
i
7 I
3
a
3
d. S perry Rand. Sperry Rand came out with new products
in short order after the announcement of System/360. Two weeks
after the 360 announcement, Univac management met to consider the
Univac Product Line Strategy. They decided to enhance and expand
the 1050 program to provide a compatible line of systems from the
1004 through the 1050 Mod V. (DX 14, p. 1.) Learson reported in
July that Sperry was "announcing new models of 1050 and 1004 where
the price/performance ratio is not following the historical trend in
the original announcement, so they are, in effect, using this as a
method of price reduction". (DX 1525.) Univac management also
decided to extend the 1107 program to the 1108 and 1109, which were
to be program compatible upwards with the 110 7, for large scale
:j users. (DX 14, p. 1.) In mid-1964 Sperry Rand announced its 1108
.i
;| at a price which Withington described as "impressive when compared to
■that of the System 360". (PX 4829, p. 20.) Withington wrote that,
"^ I
(in terms of price/performance, "IBM's initial offerings in the 360
•6 :!
:| line were inferior to it". (PX 48 30, p. 22; see below, no. 477-30.)
: .7 ;
:} (We shall see later IBM's response to this rather quick "leapfrogging".)
• S
: By 1965, Univac ' s Product Line Task Force was contemDlating
:si|
:j the introduction of an entirely new product line in reaction to
System/360. It faced a dilemma in that two of the three models
II ] -
j under development were likely to benefit from new technological
22 ">
t developments if their development could be delayed, but waiting
23 •
I would have meant that a full family could not be announced at one
i
2± 1
jtime. (DX 16, p. 2.) Univac finally compromised and announced the
-382-
3
9
ia
11
12
13
L4
Li 9200 and 9300 (rather than an entire family). (McDonald, Tr. 3821; •
Z DX 70, p. 9.) These systems "aimed at compatibility" with 360
3 ; (Eckert, Tr. 908) but achieved it only in part. (See below, pp. 480-86.p
4*j- e. Burroughs . Burroughs also responded with a new
product introduction. In August 1964 Burroughs announced the B5500
(PX 2082, p. 95) , "a more powerful successor to the earlier B 5000",
"i|
j ;} and what was to become the first member of the 500 System family.
(PX 4829, p. 22.) Withington described the B 5000 family as "incor-
porat[ing] very advanced design features, facilitating the use of
compilers and executive programs", but he concluded that "Burroughs
apparently has not attempted to answer the System 360 across the
board". ( Id. ) By 1966 Burroughs had turned the 500 family into "a
major new product line" (PX 4832, p. 21), adding the B 6500, 2500
and 3500 to the 5500 and the very large (and never delivered) 8500.
( Id. ; DX 10262, p. 8; see below, pp. 644-50.)
f. Honevwell. After the 360 announcement, Honeywell took
IS, '
■ its successful 200 system and turned it into a compatible "family of
17 ||
[computer systems": the 120, the 1200, the 2200, the 4200 and the
IS :
8200. (DX 13849, n. 27; see below, on. 619-29.) Honevwell also
19 |
j abandoned its attempts to develop a mass storage system after the
2Q ]
; 2311 introduction and began buying disks OEM.
21 |
j g. SDS . SDS announced successive new products beginning
22 :»
■j in 1964 with what it termed "the first computer to use monolithic
n \
[integrated circuits, the SDS 92" (DX 44, p. 5), and eventually, the
2d ]
j Sigma series, which was announced beginning in 196 6. (Palevsky, Tr.
25 j "
-383-
3223-24; see below, pp. 703-04.) A press release at announcement stated
that "Sigma . . . represents the first family of computers with an
entirely new design since the IBM 360 announcement" (DX 52, p. 1),
L.l-and, as IBM had done with 360, SDS stressed the new line's universal
i
applicability. (See below, pp. 704-05.)
j ;
i ;
T
3
2, i
<L
'A
•t
.0 '!
2 "»
... t
M
-384-
L
2
Z
3
37. IBM's Responses (1964-66). With competitors responding
rapidly to the initial System/360 announcements, IBM was soon faced
with the need to respond in turn or lose the competitive advantage it
had obtained by the introduction of System/360. It chose to respond.
IBM did so by introducing new products , improving existing
products and lowering prices. This section discusses IBM's initial
7 ;j responses,* particularly IBM's reduction of extra shift charges,
improvement of memory speeds , announcement of improved tapes and
disks , introduction of the Model 20 , and development of the Models
44, 67 and 90.
a. Reduction of Extra Shift Usage Charges . At the time of
the System/360 announcement, IBM was charging its rental customers a
flat rate for 176 hours of computer use per month — the Monthly Avail-
ability Charge, or MAC. For use beyond that number of hours, an
, additional use charge was billed at a rate of 40% of the per-hour MAC
rates. (DX 14295, p. 44.)
One of the ways that competitors responded to 360 was by
j reducing or eliminating charges to customers for using machines on
i
j extra shifts. An IBM Wins and Loss Report for June 1964 cited "erosion
is]
; [ of extra shift" as one of the most significant aspects of competitive
2Q .
:i announcements since System/360. (DX 13824, p. 2.) On July 29, 1964,
21 'j .
•i Learson wrote that GE had reduced its extra shift on the 400 line to
22 >
— 4
3
9
10
il
12
13 •
14
15
IS
17 ;
is
i * IBM's planning for its future products is discussed later. (See
Z - \ below pp. 378-922. )
25 ■
t
i -335-
Li 10% and was offering its newly introduced 635, "a competitor of the
*! 7094" , with no extra shift charge at all. (DX 1525.)
i
I i IBM reduced its additional use charge from 40% to 30% on
Vil- August 11, 1964, effective retroactively to July 1. (DX 13823.)
It was not enough, and IBM received pressure for additional reductions.
5 ■! On August 13, 1964, Evans and others in IBM were notified by DSD's
'
3 :
9
a
1 !
7 i| Advanced Systems Group that:
"We are currently facing severe competition in the
. medium and large scale scientific areas from such machines as
the GE 625, GE 635, PDP-6 , etc. A goodly part of this problem
is due to our additional use charges. GE, particularly, is
offering their 600 series on a 24- hour basis. Even in cases
where we are price competitive on a single shift basis, we
rapidly become non-competitive when additional use is involved.
The 30% extra shift charge is good but not nearly enough."
(DX 13640, p. 1.)
il
In addition, IBM was losing orders to the Honeywell 200,
particularly at service bureaus. In October DPD "fought" for a
reduction in extra use charges to 10 percent, this being, as Cary
wrote to T. J. Watson, Jr., in the beginning of December, one of "the
instances where we have 'screamed' for action". (PX 1265, pp. 2, 4.)
On October 14, 1964, IBM announced a further reduction in its
extra shift charge for System/360 to 10%. (DX 14134.)
| b. Memory Improvements . Within two months after 360 was
i announced it became clear that the memory speed of certain IBM systems
11 I .
\ had been surpassed by newly announced competitive machines.* A June
*1 i
.X :
- it
.0 «
.7
.S
.3 I
"i *
Even before the announcement of System/360, IBM had been seeking
li, | to include faster memories in certain IBM computers. Evans testified
J that, for the Model 60, a 3/4-microsecond memory, which he called
lz \ "startling in test", had been planned. However:
-386-
L
2
3
4-
5 i
5
7
a
ia
u
12
13
14-
15
T - 6 ,
17 I
IS •
1964 Wins & Loss Report cited "the fast memory speeds of [competitors']
new systems" as one of the "three most significant aspects of competi-
tive announcements".* In particular, the memory speeds of the
Honeywell H-2200, the NCR 315 RMC, the Univac 1103, the GE 635 and the
CDC 3800 were mentioned. (DX 13824, p. 2.)
Haddad, then Director of Technology and Engineering,
addressed this problem further in a July 28, 1964 letter to Vice
President and Group Executive Gibson:
"I am becoming increasingly concerned over the possibility
that some of the 360 machines will be technically obsolete
before they are delivered. With the recent round of price-
cutting by some of our competitors, it is even more important
that our machines remain technically superior.
"There is obviously a strong trend toward the use of
faster memories across the board. This is exemplified by
the Univac 1108 . . . the NCR 315 . . . the CDC 3800 . . . the
H2200 . . . and the RCA 3301. . . . All of these examples
appear to give the competitor a memory speed advantage at an
equivalent 360 machine level." (DX 13825.)
The need to improve memory speed, and with it processor
price/performance, was particularly acute for the larger models of
"As we proceeded down the 360 development program, and so
| to make certain that we could deliver what we were committing,
IS ;j we decided in 1963 or early 1964 to use available memories that
I were technologies that were proven and memories that had been in
2C j production. And so instead of one Model 60 with three-quarter
'l microsecond memory, we made 2 models at that performance range. -
21 \ A Model 60 with a two microsecond main memory, and a Model 62
j with a one microsecond main memory and that's what we announced."
22 ;l (Tr. 101111.)
23 j Similarly, the Model 70 was announced with * one-microsecond main
j memory because the 3/4-microsecond main memory was not yet fully
24 1 tested. (Evans, Tr. 101112.)
.i
25 i * The other two were the "magnitude of price cuts" and the "erosion
j of extra shift". (DX 13824, p. 2.)
1 -387-
r i
3 It
:
1 1
z\
.4
5
.6 ii
.7
a
s
the 360 line, the 60, 62 and 70. An IBM "Wins and Loss Report" for
August 1964 reported that "there have been no credited orders for
Models 60, 62 and 70 since June and only a few in the uncredited
category"^ (PX.3630, p. 2.) f
Within- IBM it was believed that the Models 60, 62 and 70
compared particularly poorly with CDC's new entries. On October 19,
1964, Ralph A. Pfeiffer, Jr., then Vice President and Federal Regional
Manager for DPD, wrote to Cary comparing IBM's models to the CDC
6800. (The CDC 6800 had not yet been formally announced and was
never in fact delivered) . He stated his belief that "our model 70,
with a little less than half the performance of the 6600, rents for
approximately the same amount" and recommended "that DPD request a
100% performance improvement in the Model 70 with no increase in
rental price and not more than a 20% increase in purchase price" .
(PX 1214.) On December 1, Cary recommended that the price of the
one-microsecond memory on both the Model 70 and the Model 91 be
reduced "in order to make our bids . . . more competitive from a
price/performance standpoint". (PX 125 6) DX 14504).) Those price
j reductions were announced on December 23. (JX 38, pp. 329-30; JX 10,
I App. A, «(<[ 3,5, dd. 2-3.)
•j That day, DSD President Kennard wrote to A. K. Watson:
t
■i
22 \
i
aj
2A 1
"As you know, we have undertaken a number of actions to improve
the product line and to provide specific responses to certain
technical requirements. An example of this is time sharing. We
have redesigned the 2362 1 microsecond memory and released new
models reflecting this redesign. The net effect is a lowering
of the rental and purchase price, and lower systems prices on
S/360 Models 62, 70 and 91.
— ' '! "We have also determined, through analysis of the requirements
for peripheral I/O devices, that we could achieve substantial
-388-
L
2
3
4*i
5
a
7
a
9
10
II
12
12
14"
15
is
operating efficiencies and enhance our price performance by-
developing a new multiplexor channel. We have done this and
combined the multiplexor function with the already announced
selector channel function. The net result is reflected in a
potential systems price reduction of from $5,000 to $10,000 per
month. When the new higher speed multiplexor selector channel
(2870) is integrated into our plan, it is apparent that the 2860
price for its function had to be. re-evaluated. We have completed
this evaluation, and this has resulted in a lowering of the
purchase and rental price. This price adjustment has been
released to DPD along with the 2870 multiplexor channel."
(DX 13827.)
However, these price reductions were not enough. In
December 1964 Kennard wrote to Watson and A. L. Williams that the
performance of the 6400, as indicated by CDC, would place it between
the Models 62 and 70 while " [f]ield reports indicate a price somewhat
above our Model 50". He reported that steps were being taken to
improve the competitiveness of those machines, including an increase
in the memory speed from one microsecond to three-quarters of a
microsecond for the Models 62 and 70. This and other improvement
programs for the Model 70 and Models 60/62 were targeted for release
in January 1965. (DX 14322.)
17 : The competitive advantage of CDC 6400 and 6600 over IBM
IS ■! models continued to be a concern. On March 10, 196 5, C. B. Rogers, |
1$ ' Jr., then Director of Product Programs for DPD, wrote to Lear son: j
; !
2Q j "The CDC 6600 overpowers our 70 . . . for approximately the j
'} same rental. . . . The new entry of the CDC 6400 . . . clearly 1
21 ! out-performs our Model 62 by a factor of 2 at a substantially j
| lower price for both purchase and rental. ... It is accurate
22 'j to say we are in trouble." (PX 1389*, pp. 1-2.) j
: t j
22 .! By April 1965 IBM was ready to announce a faster memory: I
2i | the 750 nanosecond (3/4 microsecond) memory. On April 22, IBM announced!
I I
~e I the Model 65 and Model 75, each having a memory speed of three- ;
I j
1 -389-
L
2
3
5
5
quarters of a microsecond. The faster performing Model 6 5 superseded
the Models 60 and 62, and the faster performing Model 75 superseded
the Model 70.* (JX 38, p^ 393.)
c. Tape Drive Improvements. Soon after its announcement
of System/360, IBM also recognized the need to improve its peripherals
in order to maintain the superiority it had achieved in the 1950s. On
7 I August 21, 1964, the System/360 Compatability Committee reported
that because of the nature of 360, peripheral manufacturers could be
expected to market compatible replacements for IBM's peripherals:
"(1) I/O manufacturers, whether independent or divisions
of computer manufacturers, are in a position to market devices
of comparable IBM capacities at approximately 20% less price.
"(2) It appears that I/O manufacturers will attempt to
sell tape drives and terminals to System/360 customers.
a
10
12
13 .
ij "(3) There will probably be concerted activity from
T # ' competitors in marketing I/O devices on System/360 in the
~ i Federal Government." (PX 3908-A, p. 4.)
They stated that:
io if
"The heretofore heavy emphasis on processor planning as
imf \ the criterion for improved price/performance should be
u i re-oriented towards I/O developments . The across-the-
I board improvements in price/performance which will be
: ^ ij required in the 1967-68 time period will probably be
I brought about more by improved I/O capability than by CPU
*^ ;} and memory improvements. As part of the regular develop-
| ment effort, such activity will be necessary in any event
-G • to keep System/360 a viable product line . . . . " ( Id. ,
I p. 22.)
21 'I .
•i
?*?
•J * IBM did not limit its memory improvements to its larger models.
23 i on January 4, 1965, IBM announced that the memory for the Model 30 had
j been improved, from two microseconds to 1-1/2 microseconds. (See PX
24. i 1288, p. 2; PX 1637, p. 2; DX 14135.)
-390-
Technological improvement was additionally important, the
Committee reported, because
"competitors will attempt to market I/O devices, with particular
emphasis on tape drives, directly to 360 users." ( Id. , p. 24.)
Tape drives, in particular r were an area that needed improve-
ment. A presentation to the DP group staff in November 1964 by a
group headed by C. J. Bashe entitled "Group Staff Review of IBM's
Technological Position in the Marketplace" summarized IBM's position
relative to its competitors from the viewpoint of research and
development. That presentation reflected IBM's unparalleled commitment
to R&D and showed that IBM, in comparison to CDC, Burroughs, NCR and
Sperry Rand, had consistently devoted a larger portion of its revenues
; to research and development. (PX 6671, p. 5.) Nevertheless, it
showed areas in which IBM was not ahead. The report concluded:
L;
z
z
4-
T
3
10
II
12
13
*•• if "We are ahead of competition in some but not all of the
,_ ! technology areas critically important to system performance.
^ \ We do not have an unassailable position of leadership in any
! function." ( Id. , p. 27.)
! It recommended attention to "box-by-box superiority" and concluded
•' -i
j that half-inch compatible tape drives was an area in which IBM was
IS : !
:| "inferior". (Id., Dp. 15, 26-27.)
is ;j — *
i A General Managers ' meeting was scheduled by Knaplund for
ZQ !
i December 4, 1964, at which technical managers were expected to report
21 :| .
j on action plans to solve the problems in the areas in which "IBM must
22 1 - •
.1 take immediate action to attain technical superiority". One of the
23 :
■*• j
[ topics was "[a] superior performance 1/2" tape drive to be announced
j in 1965". (PX 1251 (DX 14503), p. 1) This was considered necessary
j because:
i
'i
»
;! -391-
3
3
a
3
"We're outclassed in half -inch tape and apparently can't sell
one-inch tape equipment. We need a tape drive that is superior
in performance and acceptable." ( Id. , p. 4.)
A week later, on December 11, a Peripheral Task Force
reported. This group considered the use of small systems to control
peripherals, as in tape-to-printer or card-to-tape applications.
Such applications were common, applications for the 1400 series com-
puters and it was expected that the 360 Model 20 and Model 30 would
also be used for such purposes. But the Task Force believed that IBM
had a problem and could "expect to lose approximately 500 systems {in
that application area] by the end of 1965 with the presently announced
product line", with losses expected to continue thereafter. To
minimize such losses, the Task Force recommended the announcement of
the 2400 series tape drives on the Model 20 "immediately" and stated:.
^ ;■ "Low Cost Tape should be announced on the Model 20 primarily to
satisfy the longer term problem (after 1965)."
Further :
j
,S
"Low Cost Tape on the Model 30 is required to provide more
competitively priced configurations particularly to those
customers requiring 1401 compatibility." (PX 1271, p. 3,
see pp. 6, 8, 12, 14. )
„ I IBM's fears about its lack of technological superiority in
•S \
\ tapes were made even more immediate by additional actions of its
•G !
i competitors. On December 11, 1964, the same day the Peripheral Task
•i Force issued its report, C. E. Frizzell, President of GPD, reported to
T. J. Watson, Jr., on the recent RCA Spectra 70 announcement. He
Li ;
?«;
listed among the "significant advantages" of the Spectra 70:
-392-
L
2
z
4-
?
a
7
3
ia
ii
12
13
"One- third higher speed magnetic tape drives at equivalent
rentals compared to IBM. . . . Availability of magnetic tapes on
the Model 15 gives them a magnetic tape system in a price range
where we have no current entry." (FX 1272 (DX 960), p. 2.)
Honeywell, CDC and GE tape drives were also a problem. On
December 22, T. V. Lear son wrote to T. J. Watson, Jr., concerning
"out-performed, out-priced market areas". He listed "low cost tape
systems" as one of these and stated: "This is largely the Honeywell
200 story". . He called for "[t]apes on the 360/20 [to be] immediately
announced". (PX 1288, pp. 1-2; see DX 13955, p. 4.)
IBM improved its tape drives in two steps. The first step
was the announcement of the 2415 tape drive and control unit on
April 5, 1965. (JX 38, p. 377.) The 2415, a lower cost unit for the
Models 20 and 30, solved the tape drive needs of users of those
models. The second and more important step was the announcement on
-** |j August 9, 1965, of the 2401 Models 4, 5 and 6 tape drives and control
units. These 2401s incorporated several advantages in tape technology
IS |j including: 1600 bit per inch density, phase encoding recording and
17 |[ twice the data transfer rate of IBM's earlier models. ( Id. , p.
IS ■] 484.)
19 :| For the time being, IBM appeared to have solved its problems
2G in tapes with the new 2401s and the 2415. (PX 4256; DX 13950, p. 2.)
J
21 ;( Soon, however, competition, particularly from PCMs, would push IBM to
i
22] improve its tape drives even more. (See below, pp. 886-90.)
« «•
23 s{ d. Disk Drive Improvements. As we have seen, I3M made the i
" * i
! i '
24. ; j disk drive an integral part of its System/360. (See above, pp. 323-28. )j ;
25 J The 2311 disk drive, announced as Dart of Svstem/360 (JX 38, p. 86), ;j
i * i
i ■
-393-
was "the first very reliable disk drive". (Beard, Tr. 9048.) Com-
petitors initially were unable to offer a similar product. (Case,
l\ Tr. 72744; see also Withington, Tr. 56240-41.)
M' For several years prior to introducing the 2311, IBM had
: marketed the 2302, which was a drum-like file with very high capacity.
5 ji The 2302 was larger than the 2311, but with not as much versatility as
the 2311. Soon after System/360 was announced IBM found that
"it was beginning to be apparent that customers had a far
greater need for data stored in disk drives than we had
anticipated a year or two earlier when System/360 was under
development and when the 2311 disk drive was first intro-
duced." (Case, Tr. 72742-43.)
.0 !«
Thus, IBM needed a disk drive larger than the 2311 to replace the 2302
^ : and supplement the 2311. (PX 3226A, p. 4.)
Against that background IBM introduced its 2314 for two
reasons: first, since the 2314 would be larger than the 2311 it
would "provide a better relation to competition than the 2302 files".
(Id.) Second, because the improved price/performance of the 2314
| would improve the overall system performance of 360 systems on which
3 \
"| it was used, "the 2314 was announced ... to sell more 360 systems".
:| (Id. , p. 5.)
.3 i
f
■i IBM announced the 2314 on April 22, 1965. (JX 38, p. 439.)
i The 2314 "[vjery definitely" represented an advance over prior disk
11 •!
! j drives. (McCollister, Tr. 9597.) Compared to the 2311, the 2314
| provided an increased capacity of four times per spindle (Case, Tr.
i
j 72742) , an improvement of two times in data rate ( id. ) , and the
2- "!
■I ability to operate on-line. (PX 1967 (Tr. 35690).) Beard, who then
lz \
1 worked as Chief Engineer of RCA's computer division, testified that
j -394-
while "the 2311 demonstrated the reliability" of random access devices,
"[t]he 2314 not only offered the reliability but also a practical
cost for the random access user". (Beard, Tr. 9049.)
The superiority of the 2314 provided substantial benefits
to IBM. The 2314 "turned out to be very surprising in the rate that
customers found use for it" • IBM "totally underestimated the demand
for such devices" and was "hard pressed to deliver the devices as fast
as customers were demanding them". (Case, Tr. 72743.) It also had
the desired effect on systems sales:
"The availability of the 2314 has been the catalyst to make
many systems sales for previously undeveloped application use
of computers." (PX 1967 (Tr. 35690).)
L
2
3
a
7
3
9
La
LI
IZ
13
L4-
15
L7 !|
•j "During that period the entire industry and the users began
^3 !| to appreciate the importance that disk drives were going to play
;{ in the great majority of general purpose computer systems. . . .
1$ ; I believe only IBM among the major competitors at the time
; offered an alternative between magnetic card devices and disk
20 j drives, with developments proceeding along both lines. . . .
j When it became apparent that the class of magnetic card devices
Zi | was not going to be successful in the marketplace, for reasons
of reliability, and that the disk drive was a critical product,
many of IBM's competitors were left* for awhile without a satis-
factory option." (Tr. 56240-41.)
And:
"The 2314 is an example of where the product developed a
market beyond our initial forecast expectations. Every company
should have a door-opener that beats competition — the 2314 is
such a product and will continue to be only if our pricing
policy can stand the challenge of competition." ( Id. (Tr. 35692).)
Withington echoed the advantage which the 2314 gave IBM over
its competitors : .
22 :
23 :
25
e. Introduction of the Model 20. IBM's success with its
650 and 1401 had shown that small, low cost computers were important
because they helped grow the market by permitting users who otherwise
-395-
r :
L
might have been unable to afford them to obtain computer systems. ( See
above, pp. 39-44, 141-47.) In the face of that experience, the SPREAD
Committee had recommended that IBM develop a "very small" processor,
Mf even though such a processor' might not be fully compatible with the
r
2
7
3
?
a
i
z
A:
.6 ,
,7[
.3 ■
rest of the 360 line. (DX 1404A, pp. 35-36, 69 (App. A to JX 38).)
The development of such a small processor was assigned to the World
Trade Corporation's German laboratory in Stuttgart. (Hughes. Tr.
71942-43; Knaplund, Tr» 90478.) In early 1964 that small processor
was judged "not to be as far advanced in development as the Models 30
through 70", and it was therefore not announced with the rest of the
360 line in April 1964. (Knaplund, Tr. 90489.)
The need for a low-cost computer was evident within IBM. A
document of April 15, 1964, entitled "Forecast Assumptions for the
1430N Data Processing System" (the 360/20) , stated:
"The 1430N Data Processing System will offer the
advantages of stored programming to customers and
prospects for whom mechanization of data processing
has heretofore been either impracticable or confined to
conventional punched card equipment.
"The 1430N, which will have a subset of the NPL
instruction set, will be the smallest member of the
;j System/360, and will benefit strongly from the impact
'£ | of the recent announcement of the NPL line.
V3 ! "This system will bring the world of the System/360 down to
■} the price range the small user can afford.
!T ; |
j "For the first time a new technological breakthrough,
12. \ like the one realized with SLT for the System/360, will
,j be made available at lower cost to the small customer at the
23 ; same time as to the larger user.
2- i "The 14 3 ON system offers growth within the system and
| upward growth into the System/360, Model 30." (DX 13829, p. 1.)
-396-
L
2
3
4*i
5
a
7
S
a
lq :
11
12.
13
Even though, as this suggests, the 360/20 was in large part
expected to be acquired by new users , it was anticipated that a
variety of customers would find it attractive, including:
" (1) Small companies characterized by one accounting
machine installations [sic] .
"(2) New customers in this size range*
" (3) Larger unit record customers who have not yet moved
to a system. In some cases, these customers will use multiple
1430N systems. Others will use a 1430 along with some unit
record equipment.
" (4) Users of large data processing systems who still
have a considerable amount of unit record equipment installed.
In these cases, -the 1430N would replace some or most of the
unit record equipment supporting the larger systems .
"(5) The Communications Market. This market will
be characterized by customers having a number of branch
locations requiring frequent and/or prolonged contact with
the central data processing center or among each other."
(Id. , p. 2.)
After the announcement of System/360, the need for the
Model 20 increased. On July 20, 1964, Opel wrote to Learson concerning
- 3 .'banking product deficiencies and stated that "[w]e need to have a
17 ;| more competitive response to the [Univac] 1004 and other competitive
IS :| small card processing systems". (DX 14477.) Writing after the
IS {announcement of the Model 20, Withington observed that IBM had to
20 | announce "such a computer to protect its position" from "the Univac
21 11004 and 1005, the Honeywell 120 and the GE 115". (PX 4830, p. 20.)
•i
22 i IBM announced the Model 20 on November 13, 1964. (JX 38, p.
22 ;296.) Because of the need to keep its cost down, the processor did
24. 1 not share all the features of the 360 line. First, the Model 20
25 j contained only a subset of the 360 instruction set and, hence the
i
!
i
=| -397-
'i
i
i
i
L
r
j
T\
I
Model 20 was not compatible with the rest of System 360 to the
extent that the other members of the System 360 line were compatible
with each other ► (Case, Tr. 73575; JX 38 , p. 297.) Second, the
360/20 did not use the System/360 standard interface for attaching
peripherals and instead used "native attachment". (Hughes, Tr.
71992; Case, Tr. 74085.) This was done so that IBM could offer the
360/20 at a more competitive price :
"In order to achieve [the lowest possible price] , you have to
have the lowest possible cost to the manufacturer, and in order
to achieve that it is sometimes necessary and was in the Model
20 necessary to design a special unique means of attaching disks
to the Model 20 , because using the standard interface for that
purpose on the Model 20 would have been more expensive and would
have therefore unnecessarily increased the price of the Model
20." (Case, Tr. 74085.)
£
j The Model 20 announcement stated that it was "a System/360
for card processing ... a stored program approach for smaller
business needs". (JX 38, p. 296.) As noted above, however, IBM's
forecasts for the Model 20 anticipated many potential users other
than small users. The 360/20 was in fact used in a variety of ways
7 '>! by a variety of users. For example, an IBM Competitive Daily Report
■2 i| stated that "[tjhere are about 600 Model 20' s installed with communica-
a ;|
•* j tions equipment and 700 installed in large customer accounts".
•r "' I
^ .} (PX 3773, p. 2.) And Wright testified that a sample configuration of |
1 !
" • a multiprocessor Model 67 system contemplated the use of 360/20s in \
j j
£► is connection with the Model 67 in various ways including as concentrators i
i- ] I
^ j for terminals. (Wright, Tr. 13348-49.) Similarly, DX 4851, a memo-
j i
l± "s randum on the GUIDE Project on Remote Batch Computing of February !
25 I ■ ;
i
■! -398-
L
2
111
4*
5"
7
3
10
II
12
13
14
15
Iff
17
IS
1966, contemplated a large 360 as "the central facility" of a system
in which "[t]he remote terminals may be small typewriter keyboards;
but, more likely, will be Model 20 360' s, Model 30' s, or even larger
machines with their own operating systems". (DX 4851, p. 5.) "The
bulk, of the terminals planned for use would be small computers [ , ]
mostly 360' s Model 20 or 30." ( Id. , p. 6.)
As it turned out, the 360/20 was more than merely a good
competitive response; like the 1401 in its day, it became the largest
selling of the 360 systems with more than 7400 installed in the United
States by 1970. (DX 2609B, p. 182.)
In December 1965, Withington summarized the effect of IBM's
competitive responses:
"Soon after the System 360 line of computers was
announced, it became apparent that despite the basic
soundness in the line there were a few deficiencies and weak
points. IBM, apparently desiring to establish a product position
now that will remain sound for a number of years, has moved very
vigorously to remedy the deficiencies. It has announced new
products to add to the line, improved the price-performance of
the initially announced products, and adjusted marketing policy
in certain respects.
it
;| "The 360/20 extends the line downward in price, while still
IS t retaining most of the features of a full-scale computer system.
;j Considering the appearance of the Univac 1004 and 1005, the
ZQ i Honeywell 120 and the GE 115, one had to expect IBM to announce
:\ such a computer to protect its position in a market area repre—
ZL ;j senting important dollar volume. It should be effective protec-
.j tion; the 360/20 offers very competitive price-performance
characteristics. ...
x
21* \ "The 360/65 appeared when it was clear that the initial
;j 360' s at the 'top of the line' could be bested by the competition
25 I The 360/65 cannot, at present; it offers price-performance as
good as anything on the market. . . .
't
•I -399-
"When pressed by competition, IBM has also made significant
improvements in the previously announced products — even before
I ; delivery of the first models. The 360/30 initially showed a
price-performance characteristic inferior to those of some of
Uj its competitors, so IBtf increased its speed sharply by substi-
tuting a faster memory at no increase in price. The initial
terminals and control devices for remote input-output were too
expensive, so IBM has supplemented the initial offering with a
number of lower-cost devices ► Perhaps most important overall,
IBM increased the packing density of all its magnetic tape units
j\ from 800 characters per inch to 1600, at a small increase in
•j price, by using a new recording technique. This factor is
a : important to the overall productivity of most computer installa-
tions, so the entire 360 line benefited considerably^ The
i I competitors will be able to match this improvement, but for the
time being IBM's position is improved." (PX 4830, pp. 20-21.)
1 '
i
i
S ;
7 ;
si
i
— I
La ■■'■
i
u \
I
-400-
7
a
a
10
II-
12
13
15'
lo
f. The Model 90 Program. As we have already discussed,
during the 1950s IBM undertook a number of leading edge development
projects designed to advance the computing state-of-the-art. Each of
4-j- those programs (such as the 701 , SAGE, NORC, STRETCH and various
projects for NSA*) was a response to the needs and demands of users
(predominantly government agencies) who required computing capabilities
beyond the most advanced then available. All of the projects advanced
the computing state-of-the-art and, in so doing, substantially
benefited computer users and helped serve the nation. In addition
they proved extremely valuable to IBM by serving as training grounds
for future IBM managers and engineers and proving grounds for impor-
tant new concepts that were incorporated into subsequent IBM computer
if products (See pp. 68-76, 126-35, above.) With the first STRETCH
computers commencing shipments in 1961, IBM began work on its next
"super computer". (DX 4775.)
The SPREAD Report contemplated the development of a "very
large processor" beyond that which could easily be made compatible
with the rest of the line. (Brooks, Tr. 22713-14; Knaplund, Tr. 90477-
78, 90517; DX 1404A, p. 16 (App. A. to JX 38).) Work on the "high end"
was under way even as the SPREAD Committee was meeting. As Brooks
17'
13
»i
i testified: "at any point in time there was somebody working on a
22
23
24.
machine beyond the fastest one we had; in any project there should be
somebody looking for a successor to it".' (Brooks, Tr. 22844.)
| * The NSA projects are discussed at length in the classified NSA
25 1 stipulation which is DX 3420A, at «Mf 79-86, 333-369.
i
':! -401-
C i
"In January 1961, a general review was made in IBM of the
state-of-the-art in components and organizational improvements , with
the goal of making a successor to STRETCH. ... In August 1961, the
program was designated Project X (ten times STRETCH)." (JX 10, f 5.)
A general timetable for development was decided upon, and deliveries
projected for 1966 or 1967. Responsibility for Project X was given
to the Data Systems Division in October 1961. Development of the
Project X computer, which was later redesignated "Project 604" and
which ultimately became the 360/90 program,* proceeded throughout
1961-1963. (JX 10, if 5.) The Model 90 program was an effort to
"push technology" and build "the most powerful computer" possible at
the time. (Knaplund, Tr. 90571-72; PX 1034; PX 1036; PX 1041.)
The impetus for the Model 90 program was much the same as
the impetus for IBM's earlier efforts to "stretch" the state-of-the-
art. Beginning in 1961 and carrying through the Model 90 announce-
ments in 1964 and 1965, an increasing number of "leading-edge"
- i customers requiring advanced solutions to complex computing problems
began "pressing" IBM for systems with higher performance than I3M
1
y
3
I
3
3 :j
a
then had available. (Wright, Tr. 12903-94; JX 10, 1f1f 4, 9; PX 1061.)**
J Not surprisingly, as it had in the 1950s, a good deal of this pressure
■i
•1
2:1 * The Model 90 program consisted of the System/360 Models 2092 I,
j 2092 J, 2091 and 2095. (JX 10, II 1.)
3 j
| ** Such customers included the Atomic Energy Commission (AEC)
'A i facilities, the Weather Bureau, various universities, and the National
! Security Agency (NSA) , as well as private research organizations.
15 j (JX 10,"«f«f 4, 7, 9, 13.)
'j -402-
L
2
Z.
4*
§'
S
r
a
la
a
VL
12
la
IT
IS
came from the Federal governments*
"'Because of Its need for more and more computer
capability, the government has encouraged EDP suppliers to
advance the state-of-the-art. For example, in the past 20
years, the total, computing power within AEC supported
facilities has on the average almost doubled each year.
AEC has encouraged EDP manufacturers to advance the state-
of-the-art because of its requirements for advanced
computers. Throughout its history, AEC has acquired some
of the most advanced computers available." (JX 10, fl 15;
see Knaplund, Tr. 90920-21; PX 1061; Plaintiff's Admissions,
Set IV, U 37.0, 53.0-.6, 82. Or DX 7518, Mount, pp. 63-64.)
In the climate of the early 1960s, such demands were not taken lightly.
As Dr. Robinson, IBM's Director of Scientific Computing testified:
"At that time in history, the President of the United
States and the people at large had dedicated themselves
towards a substantially higher level of scientific and
engineering and technological achievement than the country
had experienced prior to that time due to a variety of
considerations, including the Russian success in areas of
technology and science, and a national goal had been stated
relative to the need for the country to achieve great leaps
forward in various areas of science and technology."
(Tr. 23049.)
Knaplund testified that in August-September of 1963 "IBM top manage-
ment was deeply concerned that IBM's efforts had not yet developed a
competitive offering for a number of very large and influential
users , especially the federal government laboratories for atomic
is ;i energy research, weapons development, space exploration and weather
20 j research, and defense contractors to the government".** (Knaplund,
21 ■■! •
i
22. ;| * According to Knaplund, government qsers and contractors were
; j "right in the forefront" of customers who had the "largest and most
23 j demanding computational requirements and therefore needed the most
'powerful computing equipment". (Tr. 90921.)
24. 1
■j ** "Mr. T. J. Watson, Jr. , and others expressed concern that IBM
25 j was not responding adequately to the needs of the United States
■i
! -403-
L
i
?
.4
Tr. 90518-19; see Wright, Tr. 12897, 12893-94.)
Development of advanced, state-of-the-art computers was not
only in the. nation's interest, but in IBM's self-interest as well:
First, as demand for such capabilities increased., so too
did the potential business opportunity in meeting those demands.
Thus, in August 1963, T. V.. Learson wrote:
"I am informed that a machine 10 times 7090 has a market
of some 53 machines. If the market is anywhere near this
number we will be committing a very serious crime in not
moving Project X . . . at a more rapid pace". (PX 1040;
see also Norris, Tr.. 5617; Wright, Tr. 12893-94; Brooks,
Tr. 22718-19; Knaplund, Tr. 90920.)
Second , there was promotional value in being able to offer
the world's most powerful computing capabilities to solve the problems
of highly advanced users. As Wright testified:
"[I]f you could take one of those leading edge
customers, or prestige, if you want to use that term, and
get him to use a data processing system to solve a new
problem that other people had not yet solved, then
~ ( j generally many other people would follow his leadership
m ?! and use the data processing system in a similar way to
■! solve a similar problem." (Tr. 12899-900; see also Dunwell,
Tr. 85840; PX 1041; PX 1082; PX 1160.)
5
3 :
Third, the opportunity to work on projects at the techno-
logical leading edge of the industry offered a powerful incentive for
II | Government for advanced EDP systems in connection with the Government's j
j high priority defense and related programs". Thereafter, he ordered j
22 :• that IBM inquire of government users directly to make certain that i
.j their needs were being taken into account in IBM's "super computer" j
23 j (Model 90) development, and ordered acceleration of development j
efforts on a more powerful computer than even the Project X computer. j
It \ (Knaplund, Tr. 90519-20.) t
-404-
?
8
9
10
LI
L the best young talent to come to work for companies who undertook
2 •{ such projects. These projects therefore served as important training
3 •:, grounds for future employees. IBM's experience on SAGE and STRETCH
4- j had provided ample proof of the benefits to be gained in that respect.
3 if (Dunwell, Tr. 85549-50; Crago, Tr. 85979-80.)
Fourth , "super machine" development held the promise of
substantial future value which would be realized through the incor-
poration of new learning in later products. (Eckert, Tr. 836-37;
Lacey, Tr. 6657; DX 13526, Forrest, pp. 106-07.) This benefit,
although quite tangible, was difficult to quantify in advance ♦ IBM's
experience on STRETCH had shown that, although high technology pro-
y* |j jects might lose money when all the costs of research were allocated
»-• | to them,* they could still turn out to be very profitable in terms of
•* | "technological fallout". (Gibson, Tr. 22593; Case, Tr. 73606-08;
•- 1 Dunwell, Tr. 85791-94, 85880-82; Hurd, Tr. 86595-98.)** "Many of the
*\
_„ '■{ technological developments made in the STRETCH program were of sig-
Ics ;{
. | nificant benefit to other IBM programs." (JX 10, 1[ 3; Case, Tr.
j 73606-08; Dunwell, Tr. 85538-49; Hurd, Tr. 86592-93; E. Bloch, Tr.
!| 91485-89; DX 3171; DX 8923.)/ Thus, T. J. Watson, Jr., writing to IBM
13 ;i
fKf^ ill ■ I i I
I * One of the problems in trying to evaluate the profitability of a
21 i program like STRETCH is that the value of technological fallout was
■j not credited to the program nor were the costs allocated to the
22 "\ benefited products under IBM's internal method of cost allocation.
| (Knaplund, Tr. 90526-27; JX 10, 113.) That value had to be taken
23 j into account in deciding whether to embark on a like program.
I
2X | ** That view was held outside of IBM as well. (See DX 5423,
■j Smagorinsky, o. 94; DX 13526, Forrest, pp. 106-07.)
25 -i
!• / In a letter written to Thomas J. Watson, Jr., on April 8, 1964,
| Stephen Dunwell, who had been Manager of Project STRETCH, called
.| System/360 the "image of STRETCH" because of all the 360 features
| which first appeared in STRETCH. (DX 3171.)
j -405-
President A. L. Williams in May 1965, stated:
"Although four or five years ago there was some doubt
as to whether or not we should continue to try to lead
in this area because of expense and other considerations ,
at some point between two and three years ago, it became
evident that the fallout from the building of such large-
scale machines was so great as to justify their continu-
ance at almost any cost. Therefore, for the past two
years, under Vin Learson and Dick Watson, this subject has
had the highest priority, at least in the upper areas of
the management of the corporation." (PX 1469.)
There were many others within IBM who felt the same way. For example,
Dr. Gibson, then IBM Vice President and Group Executive, testified
that one reason for undertaking the Model 90 program had been that
"the designing, building and operation of such an advanced
computing system had, in the past, and was believed would
continue to provide very valuable experience in programming,
in architecture, in reliability and in technology". (Tr.
22644.)
And Dr. DeCarlo, IBM Director of Systems Research and Development,
£j[ wrote in June 1964 concerning the Model 90 program:
"We can be intuitively sure that the technological benefits
which will flow from this commitment will filter through the
5 'j rest of the product line. Surely there can be no doubt the
STRETCH program spawned highly successful financial programs".
(DX 7692, p. 3; see also McCarter, Tr. 88408; DX 1141.)
j
3
Although these reasons for embarking on the Model 90 program
antedated the announcement of the CDC 6600,** CDC's announcement
.. j ** The CDC 6600 was publicly announced in July 1962 (JX 10, 11 4)
: " ■! but was discussed with customers earlier. (Norris , Tr. 5934, 5938;
.„ 1 DX 308; DX 309; DX 310; DX 13526, Forrest, pp. 191-97, 205-06, 225-30,
— j 232-42, 245, 504-08, 570-74, 580-81.)
-406-
1
2
■5
4~
3
brought the importance of that program home to IBM management with
greater force. STRETCH had maintained IBM's lead in the large scale,
advanced computer field. (Dunwell, Tr. 85736, 35741-42; see also PX
1469.) Within. IBM., the CDC 6600 caused concern about IBM's industry
leadership in state-of-the-art computing and about the perception of
IBM's role, by its customers. In August 1963, T. J. Watson, Jr., wrote
7 || "Last week CDC had a press conference during which they
officially announced their 6600 system. I understand that
in the laboratory developing this system there are only 34
people, 'including the janitor.' . . .
3
9
ia
12
13
14
"Contrasting this modest effort with our own vast development
activities, I fail to understand why we have lost our industry
leadership position by letting someone else offer the world's
most powerful computer." (PX 1045.)
The matter of computers having very advanced capabilities
was a "top priority" among the subjects discussed at the September 5,
1963, IBM Executive Conference in Jenny Lake, Wyoming. (JX 10, tf 7.)
These discussions - included "what actions could be taken by IBM to
.j- jl catch up to and surpass CDC in the area of very high performance
;j computer systems". (Knaplund, Tr. 90519-20.) IBM Research was
i!
instructed by IBM's Chairman
18 J
20 !
21 ;|
i
22 : |
23
"to ensure that IBM does have clear leadership in the computer
field — meaning a computer which is sufficiently far ahead of
any other computer — that it will maintain that position of
leadership and prestige for at least three or four years after
announcement". (PX 1049.)*
* Watson, himself, wrote one month later:
.1 "As leader in the industry, I don't see how we can afford any
24- "} other position than having the most powerful machine on the
| market. . . . [W]e should promptly commit ourselves to a
25 I machine of sufficient power so that our leadership will be
I unquestioned". (PX 1051.)
-407-
I
DSD was instructed to move ahead "as fast as possible" with Project X
(which was already planned to have twice the capability of the CDC
6600) and Research was instructed to accelerate its work toward a
machine with ten times the capability of Project X* (JX 10, If 7; see
also Knaplund, Tr. 90520; PX 1021; PX 1036; PX 1041.)
* ;i While there were reasons independent of CDC for undertaking
r i the Model 90 program, CDC r s growing success spurred IBM to advance
J ; the pace of the program by increasing the time and resources allocated
?! to it. (JX 10, 1f 8; PX 1021; PX 1041; PX 1082; PX 1204.) Neverthe-
2 ; less, the Model 90 was not announced with the rest of the 360 Series
i
1 ; on April 7, 1964, because Paul Knaplund (who was responsible for
2. I bringing before IBM management recommendations concerning the number
a i| of processors to be announced with System/360) "did not feel that the
4. I Model 90 had progressed far enough to warrant a general announce-
ment". (Knaplund, Tr. 90520-21; see also DX 9080.)* The first Model
5 ;| 90s — the 2092 I and 2092 J — were announced on August 17, 1964.**
7 :
3
:»
..1
»i
!i ) * Customers were informed, however, "that the Model 90 dev«»1nnm*»nt
"~ j effort was under way. That information was supplied in a footnote to
y? '; the System 360 announcement". (Knaplund, Tr. 90521; JX 10, «f 1.)
it \
:{
55 !
** No Model 92s were ever delivered. It was superseded by the 91
and 95, which had improved memories. (JX 10, *f 31.)
-408-
L
Z
111
i
i
5- !
t
I
« I
7
3
2
10
11.
I
12. i
i
13 '■
i
14- 1'
The Model 2091 was announced in November 1964.* (JX 10, \\ 1.) Each
of the Model 90 systems delivered to customers performed well and to
customers' satisfaction and passed acceptance tests imposed by the
government where such testing was performed- (McCarter, Tr. 88413;
DX 3162, DX 3167, DX 3224, DX 3266.)
The first Model 90 computer was delivered nine months late
because "IBM encountered unexpected, substantial and critical problems
in the Model 90 circuitry (ACPX) in 1965". These circuit problems
were "a major reason for the slippages in the delivery of the Model
90 computers." (JX 10, 1f 30; see E. Bloch, Tr. 91940-43; JX 10,
% 33.)** The principal problem, known as the "cracked stripe problem"
* Product Test non-supported these announcements because it could
not perform its standard type "announcement testing". (JX 10, 1M( 17,
21; PX 1177.) McCarter explained:
"To do this [to undertake the Model 90 program] it was
necessary to work closely with customers to understand
their needs. This requirement for customer involvement
•* !• meant that public disclosure of intention and negotiation
^ ; with individual customers had to precede the development
17 ;| of a product to a level where Product Test could conduct
,[• announcement testing. Hence, the Model 90 program was
IS ;! not susceptible to the kind of product testing applied to
other parts of System 360." (McCarter, Tr. 88409.)
* «• t
TO !
j According to Knaplund, because of the "very advanced technological i
2Q j nature of the program" IBM management placed "primary reliance" on the j
I judgments of IBM's top technical people in proceeding with the j
21 ;! announcements. (Knaplund, Tr. 90523-24.) After deciding not to j
j announce in April 1964 but before deciding to do so in August 1964, j
22 "» IBM management had already received information from the National i
j Security Agency that the (ACPX) ASLT circuitry on which the Model 90s j
22 \ depended was feasible. (See the classified NSA Stipulation, DX 3420A, I
j u 387-415, especially <Mf 403, 411-415.) |
24 1 S
j ** Advanced computers have frequently been delayed because of j
25 ! unforeseen problems. (JX 10, II 34.)
1 !
S i
1 !
] -409- ;
could not have been foreseen because it appeared only when a
sufficiently large number of components had been put together in an
operating machine. (Gibson, Tr. 22640-41; E. Bloch, Tr. 91940.)
• That problem was discovered much earlier than it would otherwise have
been because of the high current densities in the Model 90 circuits.
As a result IBM was able to correct the problem on the rest of the
360 line before most had been built and to inform the rest of the
industry about the problem- before they ran into similar difficulties.
(Case, Tr. 73594-95.)
Discovery and solution of the cracked stripe problem was an
example of the kind of technological fallout expected from the Model
90 program. As that program proceeded, additional fallout resulted
from developments in
- thin film technology (Gomory, Tr. 98273-75; JX 10,
fl 32; DX 3164) ;
!j - monolithic circuitry (Case, Tr. 73593; JX 10, «f 32;
DX 3164);
transistor technology (JX 10, 1f 32);
- packaging technology (Case, Tr. 73593; JX 10, 11 32; DX 3164);
jl - interconnection technology (DX 3164) ;
: ! - memory technology (PX 3050; DX 3164); and
- :! -
i - machine organization (Case, Tr. 73593; PX 3050) .
I :!
| Although the anticipated technological fallout from the
3 i
| Model 90 program was realized, the 90 series did not fare well
- S
;! competitively. Only 15 Model 91s were manufactured (four for internal
-410-
L
2
3
4-
?
a
7
3
9
10
11 !
12 I
13 i
IS
17 i
IS ;
use) and two Model 95s were manufactured "specially for NASA".*
(JX 10, If 35.)
19 i * By contrast, "CDC manufactured 94 Model 6600/6700 computers and
:| 121 additional 6000 Series computers". (JX 10, % \ 36.) Mr. Norris,
2G | Chairman of CDC, called the 6600 "particularly" successful. (Tr.
i 5849-51.) And in 1969, CDC Vice President, J. W. Lacey, speaking to
21 ■) a CDC graduate orientation class, said that CDC was widely recognized
■} to have "a world-wide leading position in large computers" — an area
22 ■< which CDC was able to "dominate" after delivery of the 6600 in 1964.
1 (DX 438, p. 7.) According to an IBM offer of proof (DX 1185, pp. 3-4.
23 j CDC's revenues and gross profits between 1964 and 1972 from the sale
and lease of 6600s exceeded CDC's targets, and although DX 1185 was
2i '\ not received in evidence, we rely on it because it is consistent with
■I the other evidence more fully set forth below in the CDC history.
25 j
i
i
I -411-
g* The 360/44 . As discussed above, one of the risks in
providing a line of computers like System/360 intended to do all
applications equally well was that, for some applications, at least,
the machines of the family would be, less suitable for some customers
'. than competitive machines, optimized in their design for such appli-
cations. Additionally, there was a risk that not all customers
j| would be willing to accept the "overhead" associated with System/360 f s
:| highly functional systems software — that some number would attempt
to locate alternatives with less function and better price/perfor-
mance. *
For some (certainly not all) users in such areas, this
turned out to be true. Knaplund testified that:
"In the months following the System 360 announcement,
marketing personnel began to report that, although many users
found the System 360 products adequate for data acquisition and
data reduction, some felt that a general purpose processor more
tailored to those specific applications would be required. The
Data Processing Division urgently requested that the Product
Group undertake development of a system to meet these needs."
(Tr. 90539.)
The need for a competitive response became increasingly
;j apparent during the latter part of 1964 and into 1965. Learson
•| wrote to Watson in December 1964 concerning the acceptance of the
.1
'\ Models 40 and 50 in the "Intermediate Scientific Area": "Our position
i
. 'here since announcement in April, 1964 is that we have won 44, lost
■J
Li 44, and have 172 doubtful situations. CDC and SDS have a total of
i
L | * In this context price/performance means strict throughput per
j dollar leaving aside questions of function, reliability and service
r i
• j
■|
] -412-
L
Z
J
J
5
five machines which out -price, out-perform us by a good margin".
(PX 1288, p. 2.) This was just one of many cries for a competitive
answer. (See PX 3615; PX 3630; PX 3566; PX 1439A.)
In August- 1964, DSD began a program, (called the "42S")
whose objective was to develop a processor "within the general
architecture of the System 360 family" but with better price/perfor-
j lj mance than the Models 40 and 50 for "data acquisition, data reduction
and certain scientific calculations."* (Knaplund, Tr. 90539.) That
program culminated in the announcement of the 360/44 in August 1965.
(PX 158 9A.)
The Announcement Letter described the Model 44 as "a power-
ful computer . . . designed specifically for the small to medium-
sized scientific user * . . ideally suited for customers and prospects
who want raw binary speed and high throughput to solve a wide range
of scientific problems, including high speed data acquisition jobs".
(PX 1589A, p. 1; see also Knaplund, Tr. 90539.) To reduce costs and
achieve the "raw binary speed and high throughput" needed for this
"lean, hard system", some sacrifice in compatibility with the rest of
3
1Q
Li ;
12
12
L£
15
IS
it ■;
IS
A
j * In April 196 5, Knaplund wrote that, "The performance needed [in
20 i the Model 44] approaches the Model 50. The system price required is
I close to that of the Model 30". He went on to say:
21 •!
j "Wherever possible within the framework of our main thrust
22- :! price/performance curve ... we must and will bend every
'! effort to preserve complete compatibility for marketing, as well
23 | as programming reasons. But when an anomalous performer is
| required, we must be prepared at all times to offer lean, hard
24 'j systems with slight incompatibilities, if these incompatibilities
j help mitigate impact and/or cost.
1 "Such is the case with the Model 44 . . . ." (PX 1439A.)
i
:j -413-
System/360 had to be made.* (Knaplund, Tr. 90540; PX 1439A; PX 1538;
' PX 1589A, p. 1.) The required cost savings were achieved "by
; eliminating read-only storage through the utilization of hard- wired
; logic for the interpretation and execution of stored program ins true-
jtions, by reducing the number of instructions directly executed by
i
i
I this hard-wired logic, by simplifying the checking logic and by taking
advantage of lower component costs". The required performance
increase was achieved "by using hard-wired logic in place of read-
only storage and by including within the processor a single disk
storage device known as RAMKIT for program residence." (Knaplund, Tr.
90540.)
Apart from its inability to execute the complete System/360
instruction repertoire,** the 360 Model 44 was "basically the same"
as the other 360 processors. (PX 1541, p. 6; see also PX 1589A. ) A
"New Product Programs Status Report on the Model 44 Program", dated
i one month prior to the announcement, even indicated that the 44 would
5
7 1
i
i
5!
9
* The IBM Product Group Policy for Processor Architecture (released
July 30, 1964) envisaged the need for deviations from compatibility in
order to "keep pace with systems technology and market requirements".
Exceptions from the rule of compatibility were permitted only to
achieve cost or performance improvements greater than 10%. (DX 9036.)
The improvement anticipated in this instance "substantially exceeded
10%" and was therefore "consistent" with the Policy. (Knaplund,
Tr. 90540.)
2
3
4 J
i
5
** Even this difference could be eliminated, albeit at some sacri-
fice in throughput. At announcement, IBM offered as an RPQ an extendeji
instruction set package (implemented primarily by software) which
eanbled the Model 44 to execute the "full range of System 360
instructions". (Knaplund, Tr. 90541; PX 1589A, p. 1.) In 1968, an
improved version of this feature was provided. According to the
announcement letter this "Commercial Feature" offered approximately
a 20% improvement in internal performance compared to the prior
RPQ. (PX 3563A. )
-414-
1 1
2 ;
2 \
5
6
7
S
3
10
II
12
12
14
IS
17
18 »
be manufactured in the same facility as the Model 40 and that schedule j
restraints would require the substitution of Model 44s for production
of Model 40s "on a one-forgone basis." (PX 1541, pp. 4, 6.) This sub-
stitution never happened because additional manufacturing capacity
sufficient to meet the demand for both Model 40s and 44s was authorized
prior to the Model 44 announcement. (Knaplund, Tr. 90542? DX 1154;
DX 1155, p. 2.)
The Model 44 was not particularly successful. It failed by
a wide margin to meet the level of acceptance forecasted at the time
of its announcement. (PX 2163A, p. 4; PX 2419, pp. 6-8.) At least to
some in IBM, it appeared that this was because IBM had learned to meet
customer needs generally, but had not successfully learned to specialize
within that talent. Thus, Opel, who at the time was IBM Vice President
and Assistant Group Executive, Plans and Controls, wrote in 1967:
"Why has this happened? Are we unable to plan, build, and market a
specialized machine?" (PX 1974.) And again, in August of that year:
j "It seems to me that when we specify a product for a limited market,' it
i fails. Perhaps this is due to the way we sell or, perhaps it is due
20
to the realities of market acceptance. I'm not sure which." (PX 2099;
see also PX 3555. )
1 In part, however, the Model 44 was unsuccessful because it
21 j
j was relatively quickly outperformed by later systems of competitors.
LL ..»
t 3y the end of 1967, at least some in IBM believed that "hardware per-
ii j
I formance was excellent at announcement time, but recent competitive
2d 1
j announcements have now bypassed the Model 44". (PX 2125, p. 48.) That
25 j
I -415-
1
2
3
4
e ;
7
8
9
LO
LI
12
13
L4
15
ii
16 Ii
»i
ji
18 I
19 j
20 |
21
22
23 i
i|
24 ||
ii
25 Ii
situation continued to worsen so that by 1970 one group in the
company wrote: "As a result of being consistently outperformed by
:the XDS Sigma 5, PDP 10 or CDC 3300, the Model 44 is seldom proposed,"
j (PX 2567, p. 93; see PX 2871Av)
-416-
z
3
4-
?
7
a
10
LI
12
12
14
15
17
h • The 360/67. The SPREAD Report called for the New-
Product Line to be communications -oriented, multiprogramming sys-
tems that would be capable of performing time sharing.* (Brooks,
• Tr. 2285*-60; DX 140-4A r pp. 12, 18, 19, 24, 26, 33 (App. A to JX 38).)
That objective was met, and 360 as announced included time-sharing
capability.** (Brooks, Tr. 22859-60; Knaplund, Tr. 90532-33.)
However, a number of highly sophisticated customers with advanced
requirements rejected 360 f s time-sharing approach and demanded
* Time sharing refers to "the use of a computer by many people at
once with each user having the illusion that he is the sole user of
the computer". (Perlis, Tr. 1862-63; JX 1, p. 115.)
** This was not the beginning of IBM's involvement in time sharing.
IBM participated in a number of time-sharing development efforts
before System/360 was ever announced. For example:
— Both SAGE and SABRE were rudimentary time-sharing systems.
(Wright, Tr. 13664-65; Crago, Tr. 85975-76.)
In 1960-61, Dr. Gerrit Blauuw of IBM designed a "dynamic
address translation unit" which was a predecessor for the
dynamic relocation hardware (Blauuw Box) used in the Model
67. (Brooks, Tr. 22866; Wright, Tr. 13332.)
j — In the "early sixties", IBM developed a system that would
IS ;} execute FORTRAN programs interactively and edit them — one j
: j of the "important efforts" in adapting a batch processing !
13 | language to time sharing. (Perlis, Tr. 2042-43.) !
J !
20 j — In 1963 MIT, working with I3M, implemented CTSS (which i
} Perlis called "the first example ... of a general purpose j
21 i| - time sharing system") on IBM 7090 series systems. (Perlis. j
I Tr. 1881; see also Brooks, Tr. 22739-40; Morse, Tr. 30986.)
22 : | CTSS was described by Perlis as, a "creative masterpiece". |
j (PX 299.) !
22 :} i
[Additional time-sharing work, including work on the design of reloca-
21 I tion hardware, was ongoing in various IBM research labs. (See Wright,
[Tr. 13325-28; Knaolund, Tr. 90533; DX 4823.) i
IS ! ■ ;
j i
i :
■i !
■j -417-
» !
M
.4-
time- sharing facilities not available with System/360, specifically
dynamic relocation hardware.*
In. early 1964, Project MAC*-* at MIT sought proposals for
the development of "an extremely advanced timesharing system", (Weil,.
Tr^ 7108.) IBM. bid a multicomputer configuration of a System/360
Model 50; CDC bid a 6600; RCA bid its 3301; GE bid a 635; and Univac
bid "a complex multiprocessor system" then being designed for a
classified military weapons system. Digital Equipment Corporation
bid "a multiprocessor version of its PDP 6 computer" and was "in
among the finishers". The winner was General Electric and, in addi-
tion, "a $1 million PDP-6 was purchased by MAC as a peripheral pro-
cessor". (PX 2961, pp. 1, 3, emphasis omitted.) GE won with a
- II "modified" version of the 635 and "proposed working jointly with [MIT]
in the development of the software that would reside on that hardware".
(Weil, Tr. 7111-12 J MIT "had determined that System 360 would not
^ It satisfy their needs and that they would accept only a system incor-
• porating some form of dynamic relocation hardware". (Knaplund,
I Tr. 90533.)
.3 ]
| * Dynamic relocation hardware provided a "means for interrupting a
^ :j program at an arbitrary point, moving it out of core, proceeding with
j the interruption, bringing the interrupted program back into memory .
II ! at a new location, and starting it again". (PX 1194A, p. 3; see also
] Weil, Tr. 7287-88; Wright, Tr. 13331-32; Knaplund, Tr. 90532-33.)
iZ '.!
■j ** project MAC was an advanced research project in time sharing
2 t funded by the Advanced Research Projects Agency (ARPA) of the
Department of Defense. (Wright, Tr. 13288-89; Weil, Tr. 7111;
Zl 1 DX 5613, pp. 2-3.)
i
-418-
L
2
1
4-
5
e
7
a
9-
10
11
12
13
14-!
15
IS ;
17 !
IS !
Shortly thereafter, Bell Labs also ordered a time-sharing
system from GE. (Weil, Tr. 7116-17.) Rooney testified that, after
the MXT and Bell Labs orders, within IBM. "there was a great deal of
talk about the need for such a system in our line". (Tr. 11747.)
In. mid-August 1964, IBM formed the Time Shared Task Force
"to develop an IBM plan for time shared systems . . . because
of the loss of the MAC account at MIT and other critical
customer situations in the area of real time, time shared
systems requirements" . (PX 3502, p. 1.)
The task force was comprised of individuals in IBM "most knowledge-
able" about remote computing and time sharing, who in turn scheduled
meetings with a number of the leading experts in the field such as
Professor F. J, Corbato (of Project MAC), Dr. J. C. R. Licklider
(of ARPA) , Mr. J. Schwartz (of SDC) and Dr. B. Galler (of the
University of Michigan). (PX 3502, pp. 3-7.)
In early September, Nat Rochester, a member of the Task
Force (PX 3502, p. 2) wrote to C. H. Reynolds, the Chairman:
"System/3 60 has been almost universally rejected by the
leading time sharing investigators. Time sharing systems
are likely to render obsolete systems that are not based on
time sharing. Therefore, there is a legitimate worry that
System/3 60 may not be a resounding success unless proper
£ Q ;j steps are taken." (PX 1194A, p. 1.)
?q : j He stated that "the commonest reason the customers give for rejecting
71 j System/360 for time-sharing is that there is not adequate hardware
22 '• support for dynamic relocation", even though "dynamic relocation is
23 :| not actually required for time sharing". ( Id. , p. 3.)* IBM was being
24 i!
25 j
* This was also the view of Fred Brooks, chief architect of
System/3 60, who held the opinion that dynamic relocation hardware
j "was unnecessary for time-sharing or any other purpose". (Brooks,
I Tr. 22743.) That is why such hardware was omitted from System/360.
J (See also Knaplund, Tr/ 90532-33 . )
j -419-
19 i
20 |
told that
"customers want dynamic relocation. It may be unnecessary and
undesirable but we have not yet proved that this is so. The
technical situation is very unclear and is changing rapidly on
a month by month basis as technology advances." ( Id. , p. 4.)
Reviewing the "rejection of System/360" by those desiring time
sharing, Rochester concluded:
"There is much more at stake than these few prestige accounts.
What is at stake is essentially all computing business, scienti-
fic and commercial .... we may find eventually that many of
the best programmers will refuse to work at an installation
that does not offer timesharing or offers inferior time sharing ►"
(PX 119 4A, pp. 2-3.)
He recommended that IBM "proceed with the design, construction and
release of an advanced timesharing system," and that the work be done
in public "so as to benefit from external criticism and so as to have
a favorable sales effect". ( Id. , p. 1.)
Two days later, the Research group of the Task Force reached
the same conclusion: "System/360 has been rejected or is about to be
rejected by many of the important large-scale scientific users who are
pioneering novel ways of using computers such as the 'computer util-
ity* . This has been accompanied by a shift of attention to compe-
titive equipment like the GE 635." (PX 2811, p. 1.) They also
believed there was "a great deal more at stake":
: { "The earlier concept of 'time-sharing' has now naturally led
2 ~ !{ to the 'computing utility' concept. This means that computing
;! capacity should be available right at the working place of the
22 J computer user by means of a terminal linked to a powerful
:l central computer. . . . There is a very strong probability
" j that the 'computing utility' will be the way of all scientific j
j computing in a few years, and a good possibility that it will j
2^ ; capture a substantial part of the commercial market as well. j
if IBM cannot afford to overlook a development of this scope. j
25 ; We, are currently in danger of losing all contact with the ;
leading developers of this concept." ( Id. , pp. 2-3.) j
i
]
-420- '!
L
2
Z
S
3
10
11
12
13
14.
15
16
17
13
The Research group recommended that "if IBM is to keep its
present competitive position in the marketplace", IBM must
(a) * 'immediately undertake a long range study of the •
'computer utility* concept";
(b) "undertake an immediate all-out educational effort
to explain the capabilities of System/360 and its operating
system for the multiprogramming applications involved in most
teleprocessing and communications -oriented systems (where
dynamic relocation is not required)";
(c) "undertake a project with the objective of offering
and supporting a complex 'utility type' system requiring
multiprogramming, multiprocessing and time-sharing with
System/3 60"; and
(d) "to make this intent clear", announce "a multiplexor
channel and hardware- aided dynamic core relocation capability
at once for Models 60, 62 and 70". (PX 2811, pp. 4-5, emphasis
in original. )
It was thought that only by implementing these recommendations would
£9 1 IBM be able to retain its "position of leadership which threatens
20 j to slip from us as a result of the independent development of the
21 ij utility concept to which we have only belatedly directed our atten-
22'! tion". (PX 2811, p. 7.)
23 ! In mid-September 1964, IBM's Scientific Computing Department
■I
24, 'i reported on "remote scientific computing" to the Task Force:
j
25 1 "There exists in the market place a set of key leader
"""* j accounts representative of the scientific market segment.
;l
i
1 -421-
L
2
3
3
i
7 I
3 ;
s i
LQ I
i
LI I
13 \
14;
15 !
These accounts are invariably the innovative and exoerimentally-
oriented accounts* They are the industry's spokesmen on the
advanced state of the art in computing. They materially
affect computer acquisition decisions in a variety of smaller
establishments — both scientific and commercially oriented. . .
£j "... In general ,• the accounts in the set number over one
hundred. They consist primarily of AEC computing laboratories,
large University laboratories, large research laboratories of
industrial companies, the independent non-profit laboratories,
and certain aerospace establishments.
"Today, a subset of this market, led by key university and
certain closely related laboratories, has taken a fancy to the
so-called area of remote computing. . . .
"IBM's posture has been one of silence. In the remote
scientific area we have been at a severe disadvantage because
12, I we have not made available sufficient information regarding
our operating system for 360. It has hot been stated to what
degree the operating system will support, time-sharing. We have
also not stated what additional support, if any, will be avail-
able for time-sharing.
"Our time-sharing prospects require responses to the
|| specific functions they have posed as requirements. The
1$ \ balance of the remote scientific community needs to know our
:t responses in this regard as well as more detailed information
17 '■ about operating System/360.
ig • "Certain accounts have already been lost. A small set of
:j key accounts are right now in the process of evaluations leading I
£3 to computer acquisition decisions. For every such case, deci-
; i sions disadvantageous to IBM appear to be in the offering. In 1
?g J quantity, such losses do not appear to be large. In quality, j
■j they will have a tremendous impact upon a very large market j
?-T •! segment. ... f
j I
1
22 : ; "If we do not respond on the time-sharing requirements in !
;j the near future, the time-sharing market will be largely lost I
?~ :| to GE who has responded to this requirement. A large part of j
""""' i the balance of the remote scientific market will also be in i
24 ! i
25 \ \
■ ]
j !
' i
-422- j
i 1 '•
; i
! t
i !
; i i
'■ l
I
3
9
10
11
12
13
14-
jeopardy. ..." (PX 2964A, pp. 4-6.)*
The report foresaw that the competitive threat would not
be limited to GE:
"We can expect similar emphasis on time-sharing system
design from the other competitors. The experience of Burroughs
with the D 825 and of Remington Rand with the M 490, 1218, and
other special forms of real-time computers designed primarily
for the military, have provided them with the experience
necessary to develop well-honed second generation systems
designed for general-purpose use. CDC has also had experience
in the design of real-time systems. Furthermore, the system
study efforts being conducted by CDC and ITEK at McDonnell
Aircraft, General Dynamics and Lockheed, in the area of
computer-aided design, will ultimately result in the announce-
ment [sic] generally marketable equipment to compete with Alpine
and, to a broader extent, the 2250 and its successors. The
Digital Equipment Corporation is actively marketing the PDP-6
as a time-sharing system at extremely competitive prices.
Although no real manifestation of intent has been made by
RCA and Honeywell, the ultimate gravitation of the market
toward general-purpose time-sharing systems will encourage
all manufacturers to develop a product and support plan.
"... The growing emphasis in the scientific and engineer-
ing market must ultimately effect [sic] the system selection
process among so-called commercial users. ...
£3 i
* As we explain elsewhere, GE was at this time a corporation with
I corporate-wide annual revenues in excess of $5 billion — a
-^ | "sleeping giant" to be sure, but one with the resources and techno-
logical capability to become a major force in the computer industry.
(See below, pp. 483-90.)
±z
-423-
- a
"The advent of cost-justif ied, time-sharing business on
17 ; centrally located systems should have an explosive effect on |
| the service bureau business. This business is characterized j
XS ! today by the presence of a great many users located remotely l
ij from central facility. To some extent, the current business |
?g ;j in service bureaus is limited by turnaround time. Most service j
;[ bureau customers who install their own equipment do so because j
2Q \ of the delays introduced by access to a centralized location j
~] and service." ( Id. , pp. R28-29.) I
2i i . i
i !
?? '! . !
In sum, the whole market, in all its dimensions, would be affected
by the need for advanced time-sharing capability.*
The conclusions- of the Task Force were buttressed by
■ feedback from the marketplace. ' Por example, Hart testified that
" through the joint study activity which General Motors Research had
with IBM,
"we vigorously provided input to them about what our needs
were and the importance of doing this, job right, and what
we believed was the right way to do it. . . . [w]e went
to meetings and presented our case, and we, I suspect, did
it loudly and with great conviction. Because, if we were
going to provide a suitable time sharing environment to
support our graphic consoles, then we needed to have certain
capabilities available in order to be able to do that ade-
quately." (Tr. 80278-79.)
In late 1964 or early 1965, Dr. Ivan Sutherland, Director
j of Information Processing Techniques for ARPA, contacted V. 0. Wright
i of IBM "eight to twelve" times to discuss the topic of time sharing:
"He spoke words of encouragement, encouragement in the
* The importance of time sharing to the computer market as a
i; whole was perceived outside of IBM as well. For example, Project
■ MAC and GE both believed that the computer field would evolve toward
: "an information utility". (Weil, Tr. 7116.) Various members of the
'•| computer field within the ARPA network (such as MIT, Stanford, Stanford
'i Research Institute, Lincoln Laboratories, SDC, Rand Corporation and
; i the University of California at Santa Barbara) believed that time
* sharing "was important" and should be pursued. (Perlis, Tr. 1968-69,
ij 2043-44, 2054-55.) In 1965, Withington wrote that the "market for
! time-shared computer systems and applications" was "large and growing",
j He predicted that, within five years, such systems would represent a
;j "significant part of the total computer market. In fact . . . the
[great bulk of the computer market. ..." (PX 4830, p. 14.)
-424-
L
2
1
4
5
S
7 |
3 |
i
* I
la I
11 1
12 |
13 •
14
fact that he believed that IBM should pursue development
of the timesharing concept in products and software as
a matter of not only great importance to the United
States government, but also of great importance to
IBM and he simply encouraged and wanted to be kept
aware, sort of as an insider, of how things were
going on the project.
"[I]t was my understanding that his interests were
the fact that he believed providing timesharing facilities
to the Department of Defense contractors in design of new
weapon systems, and use in other things, including health
systems and so on, would, in fact, foster the use of compu-
ters, but more importantly from his standpoint, would assist
in the solving of problems that these people in their research
and development activity were confronted with and the use of
computers would facilitate the solution of those problems at
a more rapid rate and therefore accelerate the advancement
of technology.
"[I]t was clear that he felt that two large companies,
such as GE and IBM, pursuing developments in time sharing,
was beneficial to the government, was beneficial to industry
and, therefore, that he thought that was a good situation."
(Wright, Tr. 13287-92.)*
IBM marketing people, too, were "raising an increasing
amount of clamour, putting an increasing amount of pressure on the
* Sutherland wrote the following on September 4, 1964, shortly
after GE had been selected over IBM for Project MAC:
16 a
17
IS .
;j "Project MAC'S decision in favor of G.E. has generated
jg j a very healthy spirit of competition between MIT/GE and
■j IBM. In effect, Project MAC has stated publicly that the
2Q | IBM product is inadequate and that MIT/G.E. can do better.
) MIT/G.E. must produce the best system they can in order
2i t to make good their claim. IBM must expend its best effort
j to show that its product can serve the needs of time-sharing
22 i l n fact, IBM has been slow in responding to the needs of
: interactive computer users; now we can expect IBM to show
23 1 more interest in this field. Competition between IBM and
! MIT/G.E. is a good thing; it will stimulate rapid progress
24, j in the time-shared use of computers.
-425-
L j marketing management of IBM" to provide "a product response that
2 ; would let us be more responsive to our customers ' requirements
Z\ and to our customers' demand". (Wright, Tr. 12799.) By November
i
4- j 1964 Wright in the GEM region and others within IBM became concerned
5 j. that the time-sharing movement would build "to a great ground swell" ,
i
5 I "impact" IBM's installed base of equipment and result in "a great
7 I deal of churning of the installed base, that is, the return of
3 j products that IBM had installed because of the requirement for a
g. | new capability in a computing system". (Wright, Tr. 12802-07.)
i
IQ \ According to Wright, he and his marketing colleagues were "trying
i
H S to make sure that IBM was the leading producer, vendor, for data
12 [ processing equipment and . . . that IBM did not fall behind".
12 I ( Tr - 12807.) A response was needed "to the customers who were
j
14 ! pushing us very hard to provide a product answer to their recuire-
i
75 i ments." (Wright, Tr. 12807.) It was clear that others would pro-
i
iS ! vide that response if IBM did not. (Wright, Tr. 12843-45; PX 2964A,
17 | pp. R28-29.)
,« j One of the catalysts for such response was the Lincoln
,q '! Labs Request for Proposal which came in November 1964. (Wright,
: t
20 ; i
2^ ] "ARPA must support Project MAC fully. The MIT personnel
! responsible for choosing G.E. equipment have made their best
22 \ technical judgment. They are staking their professional
j reputations on their choice. In making a decision against
23 ■! IBM, they have stimulated IBM to new efforts. Were ARPA
j to reject the MIT decision, Project MAC would suffer a
- 4 | blow from which it might never recover, and IBM would
be able to relax." (DX 894.)
-426-
Tr. 12813.) At the same time, an RFP was received from the
University of Michigan for a "central, timesharing facility".
(DX 895 .) Watts Humphrey, IBM Director of Time-Sharing Systems,
wrote to Le arson on November 15, " [t]he list of accounts who have
interest in Time Sharing is growing daily. ... By the end of
the year, I expect that this number will exceed thirty." (PX 1238A,
p. 4.) Company prestige, as well as current and future business,
was on the line. (PX 1191; PX 1246A. )
The messages from the field were heard by IBM management.
Knaplund testified that reports from DPD in late fall of 1964
revealed that a number of "very influential and highly competent
users"* agreed with the MIT analysis of System/360 and viewed dynamic
relocation as being "crucially necessary" to a broad variety of new
and advanced applications — a feature that would "accelerate and
improve the efficiency of their internal system development and
t j
2 \
i
Z :
i
_ !
f !
a ;
7;
1
3 ;
9- \
i
10 l
i
!
II '
i
12 j
13 :
i
t
15 \
!
15 ;
| * These users included MIT's Lincoln Laboratory, General Motors
17 ![ Research, the University of Michigan, Carnegie Tech, Bell Labs,
[ Rand Corporation, Stanford University and Ohio State University.
IS; (Knaplund, Tr. 90534; PX 2811, p. 1; PX 1194A, pp. 2-3.) IBM
•| Chairman (then DPD President) Frank Cary testified concerning
13 ;i the Model 67 and time sharing:
2C ! "[S]ome of our very, very best customers wanted it. . . .
". . .1 can just tell you that when customers . . . like
AT&T and the Federal Government and the universities and General
Motors Research . . . ask us to respond, we certainly at
least try to respond to them. And we didn't undertake
2* i that with anv thought that we weren't going to be able
J to do it." ~(Tr. 101808-09.)
25 |
-427-
L; programming activities". (Knaplund, Tr. 90534.) Knaplund, Hume,
Z\ Learson, A. K. Watson and others concluded that "an intensive
3 ;. effort was urgently required to review the area of time-sharing
i
4 | axx< ^ develop a plan for meeting this requirement". (Knaplund,
t
5 |. Tr. 90534-35; Cary, Tr. 101808-09; PX 1246A.)
5 I In November, a group reporting directly to T. V. Learson
7 | and A. K. Watson was set up under the leadership of Watts Humphrey
i
3 i to try to respond to the time-sharing requirement. V. 0. Wright
g | (who was made Director of Time Sharing Marketing) was called to
^q ! Learson* s home on the Saturday after Thanksgiving Day in 1964 and
^ i told to begin work that afternoon. According to Wright, Learson
£2 r said that "the resources of the company were available to us for
i
13 j whatever we needed in order to move this development forward".
i
L4 J (Tr. 12793-95, 12814-15; Knaplund, Tr. 90535; PX 1318.)
I
25 ' Starting in December 1964, IBM made time-sharing proposals
i
tg \ to Lincoln Laboratory and "a limited number of other users in order
yj | to enhance our ability to learn and understand time sharing" .
1S : (Wright, Tr. 12842-43.) According to Wright:
20
21 :|
"IBM at that point in time was looking at this whole
development as sort of a learning vehicle or process, if
you would. There were a great many things about time
sharing capability in a computer facility that IBM did
not understand . . . and we went about it on the basis
that we wanted to develop a product that would satisfy
25 ;! Lincoln Laboratory and perhaps a few other selected
:; customers, and that . . . development, would be used,
then, as a learning process for IBM to understand what
really a time sharing system ought to be, what the
~ A i facilities and capabilities should be, both in hardware
z * \ and in software." (Tr. 12825-26.)
23 :
25
-428-
L ; IBM delivered its system to Lincoln Labs "four to five
2 i months later than had been originally proposed". Although it did
3; : not have all the functions originally proposed and did not perform
l
4^j as rapidly as had been anticipated, Lincoln Labs was able to use
i.
5 . J it as a time-sharing system. (Wright, Tr. 12829-33.) Wright
i .
$ ; believed, considering the fact that it was "the first of a develop-
j\ ment program", that Lincoln Labs was "reasonably satisfied with the
a I product" . They
9-
10
11
12 J
13
14
15 !
16
17
IS
19
"expressed some dissatisfaction in the beginning, and as
they continued to work with the product and we continued
to work with them in the product, they became more satisfied
and the expression of dissatisfaction was eliminated".
(Wright, Tr. 12832-33.)
After the Lincoln Labs proposal there was a "great deal
of demand" for IBM to propose similar products to others. (Wright,
Tr. 12842.) Perlis testified that he and others at Carnegie Tech
pressed IBM to provide "the same kind of time sharing service that
MAC was developing" and were telling IBM that time sharing was
"important" and ir that what MIT and General Electric had joined
together to do was the wave of the future". (Tr. 1963-69.)
Others in the ARPA community did the same. (Perlis, Tr. 2054.)
:| IBM selected certain users who were believed to have "the
20 :
__ -j capability of using a development system" and agreed to propose to
21 1
"a limited number" in order to enhance its time-sharing knowledge.
(Wright, Tr. 12842-43.) From January 1965 forward, IBM worked
with a group of customers nicknamed the "inner six "--the University
of Michigan, Lincoln Labs, Bell Labs, SDC, Carnegie Mellon University
and General Motors. These institutions were selected to act as
-429-
22
23
24
25
L
I
t
j
z
T\
3
a
1
2
2
"consultant or adviser to the group developing the 67" because
they were "the most knowledgeable and could make the greatest
contribution to [IBM.\s] designing a product that would fit the
requirements of [the] user community". (Wright, Tr. 12905-08;
see also Hart, Tr. 80293.)
Although IBM had originally intended to propose to only
six to eight customers "to enhance [its] experience base in the
i use of the product"', that number was increased "because of the
great pressure that built up in demand from users and from the
IBM marketing organization". By October 1965, 63 proposals had
been made.* (Wright, Tr. 1284 3.)
And IBM was still quite concerned that its competitors
would steal a march:
^ ;■ ■ " [A] great many users . . . felt that time sharing offered
"" : them some additional capability that they needed. . . .
.Si!
J
'}
"In some instances they would contact or write a letter
to one of the IBM top senior executives. In other instances
they would talk to their salesmen in their facilities, and
so on, wanting a proposal, wanting to understand what IBM
could do to satisfy this requirement.
■3 :J "And all during this period of time, in general, the
industry was in a state of agitation because time sharing
appeared that it might indeed be a new wave of the future
:j from the standpoint of computing facilities for a company
-G | or an institution.
:9 \
TT .'
i2 ;j * in early 1965, IBM received and responded to requests for
; j proposals from NASA, Lewis and various other government agencies
i3 .j (including certain national security agencies) . (Wright, Tr.
j 13316-24; DX 901.)
j
15 I -430-
• • •■
2 !
3
*!
si
s i
i
7\
i
si
9-
10
11
i
12 j
13 !
■i
14 |
15 !
i
is !
;j supervision. One or more people "with technical qualifications
17 »
11 examined each such proposal ... to ensure that IBM could provide
IS •'
•[ those functions" and there was a Review Committee whose approval
19 •
; was required before the proposals were submitted. (Wright, Tr.
*° ;| 13334-35.)
21 i!
;j .
00 »
;j * The Model 67 was simply a Model 65 CPU, modified by the addi-
2- jj tion of a "Blaauw Box" (relocation hardware). (Wright, Tr. 13357;
j DX 898.) The Model 67 could be run as a Model 65 and "many" Model 67
~ A ;| users did so by running OS part of the time and TSS the rest of the
jtime. (3rooks, Tr. 22760; PX 2029, p. 1.)
ji
25 j
-431-
"[TJhere was clearly .... an understanding that if
IBM for some reason did not respond to this particular
requirement of customers' need, ... it was very likely
that those customers might very well buy such capability
from somebody else.
"[T]he significance would be that IBM would lose
business and that part of the installed base that IBM
had at. that point in time would disappear." (Wright,
Tr. 12843-45.)
In March 1965, IBM announced the System/3 60 Models 64
and 66 "for limited bidding". (PX 6209.) With the availability
of improved memory for the Model 65 in April, the Models 64 and
66 were withdrawn and replaced by the Model 67,* which was also
released "for limited bidding". (PX 1427.) Wright, who was the
Director of Time Sharing Marketing from November 1964 until fall
1965, agreed that "every time sharing system proposal made by the
IBM Corporation during that time" received his close personal
L
2
3.
4~
3
7
3
9
.0
LI
L3
14
15
• *•<
±3
17
IS
IBM was "very careful to be sure that all of our customers,
the people who had orders, knew in fact the status of the program,
what might be a problem, if it existed at that time, and how we
were progressing."" Moreover, because the customers involved were
among the most sophisticated users, they "were able to understand
the technical problems associated with the development effort" .
(Wright, Tr.. 13336-37.) The customers "understood that the Model
67 was a research and development project and that things would
change as they went along", but they were "willing to compromise
on some of the things that we said would be included in the product
and give them up if we could not produce them". (Wright, Tr.
12881-84, 13359.)
The Model 67 had its special bid restrictions removed and
was announced in August 1965 for delivery in 1966, with the TSS
operating system scheduled for delivery beginning in June 19 67.
(DX 898, p. 2.)* The problems of developing TSS were substantially
greater than IBM or the customers had foreseen. (Perlis, Tr. 1981-82;
Knaplund, Tr. 90538; see also DX 13448.) Wright testified that when
i
I he left his job as Director of Time Sharing Marketing in November
19 ij
11965, he believed there was "some" risk of slippage in the software,
20 1
j but "good progress was being made in the development of TSS" and
•j
22 "
■j * Product Test issued a "formal" non-concurrence with the
23 } announcement , although it believed the program was "in good shape",
j The non-concurrence was resolved by management. (Wright, Tr.
2± U3352-54, 13667-68; Knaplund, Tr. 90536-37; McCarter, Tr. 83416-17.)
j The difficulties which the 67 eventually experienced were unrelated
25 I to Product Test's reasons for non-support. (McCarter, Tr. 83418.)
i
i
j -432-
L !
i
3 !
4*j- fact that the degree of automatic operation of the system and
3
e
7!
3
9
10.
Li
12
13
14-1
15"
IS
"the program would be accomplished ... as it was described at
the time". (Tr. 13360.) By July 1966, however, the number of lines
of TSS code had "approximately doubled", largely because of "the
particularly its ability to protect users from each other and
from system failures is a great deal more complex than had been
anticipated". However, the first release was still expected to
be "relatively solid in terms of schedule". (PX 1826, p.. 2.)
Problems continued to develop. In August 1966 IBM announced
a delay of 45 days in the release of the initial TSS package.
(PX 3471.) Further, in the fall of 1966, shortly after learning
of performance difficulties with the TSS software, IBM made calls on
its 360/67 customers to explain the situation and to inform them that
certain functions were being decommitted and schedules delayed.
(Wright, Tr. 12876-78, 13363-66; see also DX 897.) Wright testified
that everybody had been informed and understood that this might occur:
i "All the customers understood that it was a development type
17 ;j of a project, it was a development of a system that was to
t some extent breaking new ground, . . . and everybody under-
IS | stood that there might be changes. ..." (Tr. 12879, see
!j also Tr. 13364-65.)
IS I
■! Hart testified that General Motors Research was kept fully informed
20 : i
: | of the problems that IBM was having with TSS. (Tr. 80294.)
21 -I
■j In the meantime, GE was experiencing similar problems.
22 J
GE's efforts at Project MAC were aimed at developing a software
system called MULTICS, which was to be implemented on an advanced
2A i
.1
25 1
i
■i
'! -433-
version of the 635, called the 645. (Weil, Tr. 7227-28.) GE
announced the 645 to the public in the fall of 1965, when neither
the machine nor the software was in existence. (Weil, Tr.
7232-35 •) Before the end of 1966, GE withdrew the 645 from
marketing because, it
"began to realize that what we had on our hands was a
research project and not a product. ... We were attempt-
ing to do something that had never been done before, and,
in. principle, we might end up discovering that it was not
feasible* As it turned out, it was hard and slow, but
it was feasible." (Weil, Tr» 7234.) '
Weil described the GE 645 as "being in the research project stage"
until 1969 or 1970. (Tr. 7234-35.) In fact, the GE-MIT MULTICS
operating system was never delivered by GE; Honeywell, after the
merger with GE, completed development of the software three years
* !J behind the original schedule. (Weil, Tr. 7232-33; Wright, Tr.
L\ 13375-76, 13673-74.) These problems arose because "the partici-
i
? pants in the Project MAC effort underestimated the difficulty of
* ! i successfully developing MULTICS". (Weil, Tr. 7232.) As GE's Weil
testified:
!i
"The technical task that was being attempted was extremely
J sophisticated and many of the subjects were at the state
! of the art. as it was then known, and it took a long time j
j to iron out the details of implementing some of these j
! important features." (Tr. 7232-33.) j
i . . j
I , The 645 was never delivered and Project MAC received, I
j !
2 •; instead, a system designated the "636". (Wright, Tr. 13375-76.) j
3 1 Rather than providing GE with the "top-of-the-line prestige lustre" •;
i j
^ j which had been expected, the 645 provided "very little to General j
'\i !
c 1 I
~ j !
1 -434- i
I
2
3:
i
*i
i
5"!
t
1
si
t\
i
3 ;
i
9-1
I
10 ;
11 !
12 !
13 !
j
14-:
h
15 ;
Electric except a drain on its resources". (Weil, Tr, 7236.)*
IBM did not give up on TSS or the Model 67.** Release 1
of TSS was made available by IBM in October 1967. (DX 3282A.) By
April. 1969, IBM had delivered a ,r substantially improved" version of
TSS which was "considered to be an excellent software programming
system". (Wright, Tr. 12842, 13375; DX 905; see also Hart, Tr.
80296-3 00, 81961-63; Cary, Tr. 101809.)/
The Model 67 was not widely accepted, and by year end 1970,
only 52 Model 67s had been installed by customers. (DX 2609B,
p. 182;// see Cary, Tr. 101809.) However, the experience that IBM
gained with the Model 67 and TSS proved invaluable. Evans testified
that when he returned to SDD from FSD in 1969,0 he launched an effort
* IBM's difficulties with TSS and GE's with MULTICS were hardly
unique in the industry's development of large operating systems,
particularly for time sharing. GE also encountered problems with
its GECOS operating system. (Weil, Tr. 7215-19; see also Withington,
Tr. 56727-31; below, pp. 501-03.) As we discuss below, so did manv
lo :! others. (See pp. 479, 568-72.)
17 !
is ;
** The magnitude of the task was so great, however, that IBM did
consider withdrawing the Model 67 at one point, (px 1955 (DX 13866).)
:| / That view was not universally held. Per lis testified that
13 ;j "TSS is working today", but that it never delivered the "work load"
;| that Carnegie "expected that it should". (Perlis, Tr. 2118-19.)
20 :
.; // We are aware that DX 2609B is not in evidence but we rely on
2! ; | it for the number of Model 67 installations because it is a sworn
j statement by an IBM corporate officer based upon IBM's accounting
22 ■» books and records .
23 j Evans believed that his being sent to FSD was in some measure
j a punishment tor failing to have dynamic address translation hard-
24- j ware incorporated into the design of System/3 60 from the start.
1 (DX 4740: Evans, Tr. (Telex) 3950.)
25 j
=1 -435-
L
I
M ! storage capability became a staple of all 370 systems announced
to get dynamic address translation hardware, the hardware which is key
to virtual memory systems,* put into the 370 plan. (PX 2487A, p. R2;
DX 4740- Evans, Tr. (Telex) 3937-41.) Evans was successful, and virtual:
i after August 2, 1972. (Gary, Tr. 101809-10.) Moreover, virtual
memory function was incorporated in 370 "in almost exactly the same
way as the Model 67". (Case, Tr. 73403, 73612-13, 73578-79; Cary,
Tr. 101809-10; see also PX 2 4 8-7 A, p. R2; DX. 8066.) Thus, the Model
67 development produced hardware and software that became important
elements of IBM's computer systems for the next ten years.
3
7
a
?
a
z
3
A-
**
J I
.$ ;
i
21 ;j
j * Virtual memory or virtual storage is a combination of hardware
22. '} and software which allocates to the machine itself the task of moving
| data into and out of main storage from auxiliary storage. Virtual
23 j storage greatly simplifies the programmer's task because it relieves
! him from the burden of having to make sure that his data will fit
24 "i into available main memory space at all times . For programming
I purposes, virtual storage gives auxiliary storage the appearance of
25 'j being main memory. (DX 4740*- Evans, Tr. (Telex) 3943-54.)
i
•i -436-
14
15
15
1 I 33- Educational Allowances. Universities had played a key
2 | role in the beginnings of EDP in the 1940s and 1950s (" [t]h.e first
3 'computers were conceived and. built at universities » . ."(DX 5504, p..
4 | 15)), and a close working relationship had arisen, between academicians
i
5 !| and EDP manufacturers.*' Also, during the 1950s and 1960s, many
5 ! f colleges and universities, supported in part by the National Science
j jj Foundation and other government agencies, greatly expanded their
Ij
g j utilization of computers* (Plaintiff's Admissions, Set IT, If 641.2.)**
g ijThe number of campus computing centers grew from 40 in 1957 to 400 in
in !| 1964 (DX 5504, p. 15), and, as the Rosser Report (DX 5504)/ acknowl-
ii j
T<? ! * For example, as noted earlier, Eckert and Mauchly, the developers
i2 iiof ENIAC, did their early work at the Moore School of the University
13 ij of Pennsylvania. (Eckert, Tr. 712-15.) John von Neumann, whose
Ij papers contributed to the development of the modern stored program
ii concept, was a member of the Institute of Advanced Study at Princeton
land later a consultant to IBM. (Hurd, Tr. 85614, 86599-600.) Herman
'iGoldstine, one of von Neumann's closest collaborators, joined IBM
jjaround 1958 (Gomory, Tr. 98154), and became IBM's Director of Math-
ematical Science in IBM's Research Division. (JX 5, p. 57.) In more
'; recent times, Phillip McC. Morse, Director of MIT's Computation
:; Center, is a member of CDC ' s Board of Directors, and Harold Brown,
17 .(President of the California Institute of Technology,, is a member of
'i IBM's Board of Directors. (PX 5779, p. 33; Morse, Tr. 30961.)
13 ; ;
** As early as 1956, the Atomic Energy Commission was giving grants
19 iito universities in order to support the usp of computers. (DX 5424,
= Pasta, pp. 11-13.) By 1963, at least eight government agencies
20 ^contributed to -ne support of computers in colleges and universities:
.National Science Foundation, National Institutes of Health, Atomic
21 lEnergy Commission, Advanced Research Projects Agency, NASA, Air Force
Office of Scientific Research, U.S. Army Research Office and Office of
22 Naval Research. (DX 5504, p. 43.)
23 / The Rosser Report, published in 1966, was the work of an ad hoc
committee, the Committee on Uses of Computers, appointed by the
24 National Academy of Sciences. J. Barclay Rosser of the U.S. Army Math-
hematics Research Center of the University of Wisconsin chaired the Com-
25 mittee. The Report estimated that in 1964, colleges and universities
";had about $250 million worth of computer equipment installed in those
;'400 centers. Universities' annual EDP budgets were comparable to the
ijcosts of running their libraries. (DX 5504, p. 15.)
I! -437-
z
4*
5
!
t\
3
ia : .
i
12 |
13
U !
i
t
15 I
16
17 j
IS
19
20
2i ii
edged, computers were becoming more and more important on the nation's
campuses:
"Campus computers are used by an increasing number of
students either to do homework or laboratory problems, or to
learn about the design and operation of computers themselves .
»
"Campus computers,- like laboratory equipment, are needed to
do research* They increase the effectiveness of other scientific
equipment and permit many scientific studies of a scope and depth
heretofore unattainable," (DX 5504, p. 15.)
While in 1957 computer costs represented only 3% of all university
research and development costs, by 1963 the percentage had more than
tripled to 10.04%. (DX 5504; p. 66.)
But government funding was insufficient to support the
growth in computing which universities were experiencing during that
period. Computer equipment was expensive, and universities could not
afford it without additional help. In 1963, for example, colleges and
universities spent about $97 million on computers. About half of that
came from federal sources, and colleges and universities themselves
were able to pay for about 34%, a shortfall of 16% remained to be
provided from other sources. (DX 5504, pp. 18, 21.)
In order to make up that shortfall, colleges and universities
turned to computer equipment manufacturers for help. (Morse, Tr.
30965.)* The business equipment manufacturers had historically offered
22 j * DX 5462, a listing of Requests for Computing Hardware compiled
;|by the National Science Foundation, lists 366 proposals from 175
23 ^educational institutions from 19 57 to 1967 asking computer manufacturers
if or free or discounted eauipment. (DX 546 2, p. 20.)
24 ;i "
-438-
L
2
special discounts to universities,* and that practice was continued.
For example, when asked to explain why Burroughs offered educational
discounts, Macdonald testified:
"First of all, it appears that it's been an industry prac-
tice for a very long time.. .... [A] long with that, the educational
institutions appear to have grown accustomed to this practice
and remind us of it should we forget, and it is practiced by
5; our major competitors and it seems to sort of satisfy the general
social pressure that educational institutions should be treated
7 lj in a kind of special category as far as pricing is concerned."
(Tr. 6986.)
3
9
10
II
12!
14*
15
Thus, the pleas were generally successful:
"In the recent past, operating costs for computer centers have
increased too rapidly for the usual university financing. . . .
This difficulty has been partly alleviated by the generous educa-
tional contributions offered by some of the manufacturers ..."
(DX 5504, p. 20.)
Helping universities acquire and use computers was clearly in
the self-interest— or, as DeCarlo of IBM put it, "enlightened self-
interest" (DX 7514, p. 8)**-- of computer manufacturers. The use of
computers at universites was an important means of gaining the wide-
ns ; spread acceptance of the new technology. It offered the promise of
*7 'I overcoming some of the ignorance, fear and uncertainty about computers
*;l
13 ;j * For example, National Cash Register Co. offered educational
i[ discounts on cash registers and accounting machines at least as early
2CJ :| as 1929. (DX 347, p. 2.) Similarly, Raymond Macdonald, chief execu-
'i tive officer of Burroughs, testified that the educational allowance
21 I "practice was in effect when I joined the business in [the] mid-
i 1930s." (Tr. 6986.) IBM offered educational discounts in the mid-
22 '} 1930s on equipment to be used for teaching and research. (JX 28, II
11.)
j ** According to DeCarlo, "The evolving patterns of corporate support
2- i of education predicate corporate giving on the basis of enlightened
j self-interest, a concept that serves to illuminate the mutual nature
25 l of corporation-education relationships . The long range interests of
IBM and education coincide in important ways." (DX 7514, p. 8.)
-439-
by training the new generation in their use.
There were more direct potential impacts .. The infant
industry was suffering from an acute shortage of people who were
»
trained in computing,- educational, discounts would help- alleviate that
shortage „ According to the Rosser Report:
" [Educational discounts were] first instituted because the
manufacturers realized that they would have trouble selling
computers unless people capable of using them were available. To
encourage the training of such people, manufacturers gave dis-
counts to schools offering courses related to computers ; the more
courses, the greater the discount." (DX 5504, p. 44.)
Also, as more and more people became knowledgeable about computing,
additional applications for computers would inevitably be created, and
the market would grow. The Rosser Report described that phenomenon as
follows :
" [Ulniversities can draw upon the talents of their students,
1 1| the best minds of each generation, at a time when these minds are
alert, inquisitive, and full of fresh ideas. Because a university
can bring these minds into contact with the computer in an atmos-
phere conducive to research and imaginative thinking, it can
.. I stimulate bold and original ideas for improving the computer and
making better use of it. There is, therefore, great value in
supporting such activity in universities." (DX 5504, pp. 28-30.)
T \
3!
2
In addition, some people believed that computer manufacturers
il would derive a positive public relations return from an active program
ij
j in support of higher education. DeCarlo of IBM believed, for example,
i
that "beyond fulfillment of 'corporate citizenship' responsibilities
1 1
•j there is significant potential for public relations return on the
.(education support investment". (DX 7514, p. 6.)*
3 ;
i
* Some people thought that such a "public relations return" would
5 i include students who, having been trained on computer equipment of a
;j -440-
i
1
2
a
4-
5
For those reasons, IBM and other vendors offered a variety o:
support to educational institutions. IBM offered educational allow-
ances of varying percentages depending on whether the equipment was to
be used for administrative or instructional purposes, and IBM also
donated computer time to universities under circumstances that would.
6 ; ensure that the time, would be made available to a wide variety of
students** In addition, manufacturers, especially CDC, offered
research grants** (Norris, Tr. 5647) , "buybacks" of computer time/
7 I
S I
10
11
particular vendor t would later be inclined to favor that equipment.
ji (See, e.g., Hangen, Tr. 10448-49; Rooney, Tr. 11880 ("[W]e felt it was
ji advantageous to have the students in the university become acquainted
ii with computers by first utilizing RCA equipment.").) Other evidence,
^2 ji however, suggests that any such advantage was more apparent than real.
;! (See, e.g., Perlis, Tr. 2033; Morse, Tr. 30985; Andreini, Tr. 47880-
£3 |l 82.) As Wright testified:
; t
^4 !j "If you train a person on the use of a computer, he has an
easy time going to some other manufacturer's computing system and
£5 II adapting to that particular computing system. The fact that he
;j was trained on an IBM system does not lock him into an IBM system
and he is, therefore, able to handle another system." (Tr. 12910.)
16
17
13
19
20
21
* In the mid-1950s, IBM established data processing centers on the
ji campuses of MIT and UCLA on the express condition that any student fron
; any college in the Northeast could apply, for time at the MIT center,
•and that similarly any student in the West could use the UCLA facility.
:; (Hurd, Tr. 86421; see also Morse, Tr. 30965.) Almost 40 colleges and
; , universities ultimately participated at the MIT center and over 60
:! participated at UCLA. (DX 7514, p. 33.)
** Norris of CDC defined a research grant as a situation in which
"the educational institution pays the list price, but in a separate
■transaction we sponsor a particular research program at our expense".
22 ;; (Tr. 5647-48.)
2,5 '.; / Norris said that a "buyback" meant that "Control Data had the
^ right to use time on the machine for its own purpose in an amount equi-
*4 valent to that expressed in terms of the monthlv rental". (Tr.
,5988.)
25 ;
• -441-
I!
ii
ii
1 :j(Norris, Tr. 5988), and large cash contributions.* (Morse, Tr. 30980.)
Larger discounts for academic or instructional use than for
3 ! administrative use were not uncommon, and were consistent with the
4 ildesire of manufacturers to encourage the training of people knowledge-
i
:i •
5 !iable in computing. The Rosser Report observed that:
I
6 .1 "There are numerous cases where computer companies have given
ij support in the form of generous discounts on the rental or pur-
7 |l chase price; in many cases, this has been done on the condition
;| that some computer time be made available for instruction. This
3 si has ramifications similar to those arising from the NSF [National
|l Science Foundation] support with its side condition." (DX 5504,
9 j P-. 57.)
1 1
IQ IjLooking toward the future, the report argued t
!!
H Is "The computer manufacturers are gravely concerned with the ques-
tion of education. For one thing, the shortage of programmers,
^2 |j referred to earlier, has a dampening effect on sales of compu-
ij ters . If the computer manufacturers could be assured that their
^3 jl educational discounts would really support education, rather
II than being used in good part as an indirect subsidy of government
^4 !i research, they would be disposed to return to the more liberal
j discounts of earlier days."** (DX 5504, p. 58.)
15 ;
15 i
17 I
13 ;
19 '
20
21
22
In restrospect, the educational allowance plainly accom-
plished the goal of supporting the growth of the industry, as well as
* In the early 1970s, for example, CDC donated $5 million to MIT,
payable over five years. (Morse, Tr. 30980.) Morse, head of MIT's
.Computation Center, was also a director of CDC but he testified that
'there was no connection between that fact and CDC's contribution. ( Id. j
** The reference is to the Carnegie decision (PX 1088) , a ruling by
the Armed Services Board of Contract Appeals that universities had to
pass educational discounts on to the government whenever computers
acquired on such a discount were used in government-sponsored research.
The Rosser Report criticized the implications and developing repercus-
2j sions of the Carnegie decision and concluded: "In view of the pressing
need for education in connection with computers at the present time,
24 -this tendency of government officials should be reversed." (DX 5504,
p. 45.)
25
-442-
benefiting the society in general. In a draft report prepared for
the National Science Foundation, Prof. W. F. Miller of Stanford Univer-
sity concluded, that the educational allowance
"was a very important form of support in the early years. It
contributed immensely to the growth of the computing industry in
the country* The computing industry grew in its most spectacular
growth from the ground up. When the colleges and universities
began to graduate engineers, scientists, business school
graduates , etc . ,. who had been introduced to computing through
7 if introductory courses and. often had taken advanced courses in
computing, they began to introduce computer methods into their
3 if respective businesses. This in turn stimulated the great demand
for computers and the spectacular growth of the computer industry
in the early and mid-1960s. There is no doubt that the colleges
and universities who first introduced large teaching programs in
computing would not have been able to support these educational
courses on such an extensive scale without the benefit of the
[educational allowance]." (DX 5500, p. 3.)
10-
II
12
13
14-
15
IS
17
13
Similarly, the President's Science Advisory Committee,
writing in a 1966 Report titled "Computers in Higher Education",
observed that:
"Great good has been done through donated computers , obsolescent
computers , huge educational discounts , grants for the purchase of
computers and the struggles of enthusiastic men with inadequate
machines." (DX 5476, p. 18.)
Speaking of IBM's educational allowance program, Hurd tes ti-
ts il
-° it fied that educational institutions and society in general benefited
from educational allowances :
2$ ;j "First, because of that educational allowance policy they were
able to afford the installation of general purpose computer
systems, and having afforded it, they were able to support their
instructional programs , support their research programs , and as I
have indicated in my testimony, increasingly the use of general
purpose computers supported research not only in the physical
sciences but in the social sciences and in the humanities . So
I in that sense the IBM educational allowance policy contributed to
2- [ > society in general because of the research results and the
:! instructional results." (Tr. 86 715.)
i
; ! -443-
77
25 : :
3
IBM's Educational Support Programs. IBM's support of educa-
tion started with the beginnings of the company and it was originated
and directed to a large degree by Thomas J. Watson, Sr. (DX 12150,
pp. 17-18? DX 7514, p. 8.)
Watson had a very strong belief that IBM should support
educational institutions because " ' [t]here is no saturation point, in
education."* (DX 12150, p. 18.) When IBM's SSEC became operational
in 1948 (see pp. 24-25 above), Watson dedicated it to science and
IBM allowed educational and research institutions to use the machine
without charge. (Hurd, Tr. 86420.) At the dedication of the SSEC on
January 27, 1948, Watson told the assembled guests:
"It is with a mixed feeling of humility and confidence in
the future . . . that I dedicate the IBM Selective Sequence
Electronic Calculator to the use of science throughout the world."
(DX 12150, p. 32.)
Mr. Watson's strong belief regarding IBM's support of educa-
tion and the mutual benefits which would accrue was the basis for
IBM's continuing policies in this area. As Dr. Hurd testified, IBM
7 ij "hoped to benefit from the expansion in the understanding, uses and
■I
5 ;{ users of computing . . . and I believed that all suppliers of general
9 j purpose computers would benefit from its policies". (Tr. 86422.)
G .1 He went on to comment: "they [educational institutions] were able to
i
1 | afford the installation of general purpose computer systems ... to
support their instructional programs . . . research programs ... in
i
the physical sciences . . . social sciences and in the humanities". I
I
6 1 (Tr. 86715.) !
[5 j There are many examples of IBM's early support of educational !
! and academic endeavors. For example, in 19 24, Henry Wallace, who j
i !
,1 -444- !
1 I later became Vice President, used IBM punched card equipment which had
2 i been donated by IBM to Iowa State College, in order to do the research
3 :} that led to his Ph.D. degree and later to the invention of hybrid
ii
4 i| corn. (Hurd, Tr* 86712.) During the 1930s, IBM supported Dr. Wallace
5 i'JEckert's astronomical research at Columbia University and in 1945
5 .j established the Watson Laboratory at Columbia. (Hurd, Tr. 86713;
7 -iJDX 12150, pp. 11, 23.) In the 1940s, IBM supplied the machines that
:l
ft
g.'ijWere used by the Carnegie Foundation for the Advancement of Teaching
\\
q »J to evaluate test results from a nationwide survey. (Hurd, Tr. 86713.)
it
in !l In October 1955, IBM announced an educational allowance
.- j; program for the 650 computer. An allowance of 60% off the rental
.- :i price was available to educational institutions that offered courses
,3 ;jin both scientific computing and data processing. (JX 28, fl 12.)
The 60% discount provided a great benefit to universities. Perlis
testified:
14
15
15
17
"The 60 per cent discount that IBM made available to
universities opened up digital computing in the universities
in the sense that almost no university was able to afford
or at least thought they could find funds of the kind
required to establish a digital computer laboratory until
IS • that discount became available, after which there were just
a very large number of IBM computers, in particular IBM
1$ : 650' s finding their way into the universities and forming
the focus of university computer centers." (Tr. 2009.)
20
21
22
That educational allowance policy was "absolutely" one of the "principa
forces which enabled universities to become competent in computing as
soon as they did". ( Id. )
23
In May 1960, IBM announced that a 20% allowance would be
24 .
roffered on all of its EDP machines, systems and features, leased or
25
"purchased, used for administrative purposes, and that a 60% allowance
I:
I: -445-
would be available if that equipment was used for instructional pur-
poses. (JX 28, 1f 13.)*
IBM's allowance program remained relatively unchanged**
.j|- until February 1963, when IBM, abandoned the administrative use /instruc-
tional use distinction and reduced the 60% allowance to 20% on all new
orders. (JX 28, fl 17.)
When System/ 360 was announced in 1964, IBM left the percent-
age of the educational allowance unchanged (at 20%) and made it
available on System/360 equipment to colleges, universities and junior
colleges. (JX 28, 11 19.) However, it soon became apparent that
educational users who had second generation equipment installed under
the much higher 60% allowance could not afford to take advantage of
the price/performance improvement that System/360 offered because the
educational allowance on System/360 was so much lower. (See, e.g.,
DX 12435, Armstrong, pp. 80-81.)
In addition the competition for the business of educational
institutions was especially severe during the middle 1960s, and most
of IBM's competition offered high educational discounts and other
I i
I
■j special arrangements such as lucrative research contracts. (See, e.g.
iNorris, Tr. 5647.) Macdonald of Burroughs said that on occasion
a i
.j * IBM estimated that in 1962, it would grant allowances worth $24
Z i million. (DX 7514, p. 30.)
J
j | ** Junior colleges and post-high school vocational institutions
I became eligible to receive educational allowances for certain equip-
,4. Iment in 1961 and 1962 respectively. (JX 28, «Mf 15, 15.)
-446-
L
Z
3
4-
f
e
7\
3
9
ia
11;
12
14- ;
15
IS
Burroughs' discount reached as high as 50% (Tr. 7534), and Norris of
CDC testified t
"[W]hen we first started the company, as I recall, educational
discounts then were around 20 per cent. And then they increased,
and at one period X recall their getting as high as 60 per cent.'*
(Tr. 5648.) .
In fact, CDC would often offer combinations of discounts, research
contracts and buybacks in certain competitive situations. For example,
at Battelle Institute, CDC offered a 20% educational discount on a
CDC 6400, and in addition CDC gave Battelle $10,500 a month as a
"buyback" of computer time. Similarly, at New York University, CDC
offered a discount of $15,060 a month on a CDC 6600, and $14,400 as
a "buyback". At the University of Illinois, CDC offered a 1604
computer for 13 months on what CDC called a "100% Rent-free consign-
ment". (DX 278, pp. 1, 6; see also Norris, Tr. 5989-91, 5993-95.)
Plaintiff's witness Wright, a former Regional Director of
Marketing for the GEM region in IBM's Data Processing Division, was
asked whether IBM's educational allowance served any purpose from a
^ I; marketing standpoint. He responded in part:
-S || " [A] 11 vendors to some extent had educational allowances to my
;j knowledge. All of the companies I have been associated with have
-3 jj an educational allowance and this, in turn, permitted IBM to
;| compete for business in educational institutions, in addition to
2G ,| providing computering [sic] equipment at a lesser cost to those
:i educational institutions." (Tr. 12958.)
21 \ ■
\ As reported internally by IBM employees in 1964, Burroughs, CDC and
22 j
1 GE were offering a wide range of discounts, from 20% to 60%, along
i with other significant considerations such as cash grants. (PX
24. 1
:! 2963, pp. 7, 9.)
45 i
: -447-
During the period from 1965 to 1968, other computer manufac-
turers were successfully marketing computers to universities . For
example , DEC reported that its computers were used at Yale University ,
MIT, Stanford University, University of California at Berkeley, Oxford
University, Harvard University and University of Wisconsin, among
others. (DX 13846, p. 6,- DX 13847, p. 7.) SDS reported that users of
ij its computers included Johns Hopkins University, Duke University,
University of Delaware, Michigan State and UCLA. (DX 983, p. 10.)
And Hewlett-Packard reported that colleges were using its equipment.
(DX 11011, p. 10.)
As a result, IBM and its competitors frequently became
involved in highly competitive situations at universities. (See,
e.g., PX 1824 (Berkeley); PX 1468 (University of Pennsylvania);
PX 1558 (University of Colorado).)
Even if IBM had discontinued its educational allowances
altogether, it seems probable that other manufacturers would have
;j continued them nevertheless,* and it is therefore hardly surprising that
'{ to some extent IBM considered such allowances a competitive necessity.
■j As a result, in 1965, IBM raised slightly the educational allowance on
| System/360 computer equipment, and created a sliding scale of discounts
f ':
| ranging from 20% on the Model 30 CPU to 45% on the larger CPUs.
\
i :l
* When asked what Burroughs would do if IBM were forced to discon-
; j tinue its allowances, Macdonald said, "I believe we would consider it
j carefully, and were that to happen today I think we would probably
1 ) continue the practice." (Tr. 6987.) Moreover, Macdonald said that
| "I suspect that for the remainder of the industry that the practice
j : | would continue . " ( Id. )
j -448-
L
Z
Z
5
Even though that increase helped colleges and universities
to acquire the newer System/360 computers, and even though it enabled
IBM to be. more competitive, a great debate ensued, within IBM as to
whether high educational discounts were the most appropriate way for
IBM to support education.* Some favored continuing the discounts;
5 ;! some favored raising them; others favored lower ding or even eliminate
7
3
9-
a
12.
!3 '•
14
ing them.. Still others favored massive efforts in support of educa-
tion.
As examples of the differing opinions, T» V. Learson was
quoted as saying i
"We [IBM] have two objectives in this [E.A.] program: the
first to get university customers back up to paying full
rentals or as close to it as possible in the long-run; the
second, to get more revenue in the short-run, i.e., 1966."
(PX 1652, p. 1.)
Herman Goldstine, when he was Director of Scientific Development at
DPD Headquarters, observed that "the educational allowance was ori-
{[ ginally introduced by IBM as a matter of enlightened self-interest
and the expressed" intention was to further the training of young
people in the use of the computer." He went on to recommend "that we
should substitute for the EA or for most of it a cash .grants program
17 \
IS ■
is ;,
: or a value-received program. Perhaps we will always want a 5% discount
20 ':!
1
I for psychological reasons." (PX 1679, pp. 1, 2.) Armstrong took the
21 \
;] position in November 1965 that the educational allowance program be
22. 'i ■ •
\ * That debate was fueled by the Carnegie decision, referred to above,
24. ! which had the effect of passing the manufacturer's discounts on to the
(government, a result which manufacturers had not intended.
25 .1
\
!l -449-
1 left alone, i.e., unchanged from the March 1965 position. (PX 3871:
2 Armstrong, pp. 145-46; see also Wright, Tr. 12912-13; compare
3 PX 1661, p. 1.)
4 The result of the debate was a corporate decision gradually
5 to reduce the allowance to 10%. (PX 1706, p. 2; PX 1745, p. 2?
6 px 1746, p. 2.) In 1966, the educational allowances on most
7 equipment were reduced by about 10% of the price of the equipment.
8 (JX 28, 11 26) as the first phase of the larger reduction. Xn 1969,
9 the allowance was reduced to 10% as planned (JX 28, II 29), where it
10 remains today on most products.
11
12
13
14
15 I
16 S
17
18
19
20
21
22
23 |j
i;
24 !|
II
i:
25 i,
-450-
L; 39. IBM's Unbundling.
2 l a. Introduction. Before discussing IBM's unbundling in
3 ;j 1969, it is useful to review briefly the causes and effects of bun-
4.|dlingin the 1950s and 1960s. Bundling, as used in the EDP industry and
3 it with respect to IBM, is "the offering of a number of elements that are
i'
3j considered to be interrelated and necessary from a customer's point of
;(
j i( view, m the computer field, under a single pricing plan, without
detailing the pricing of the component elements themselves." (R.
Bloch, Tr. 7603-04.) The elements which were offered without a
separate price were non-hardware elements such as education, software,
i
systems design, and maintenance.* (see above, pp. 56-67.) As de-
scribed above (pp- 53, 56-67), the provision of such support services by
manufacturers greatly facilitated the marketing of their equipment to
users by reducing the users' risks in installing that new, unfamiliar,
and expensive object, the computer. (See R. Bloch, Tr. 7751-54;
3
9 .,
i
10 i
u !
12 1
T1
14;
I
I Norris, Tr. 6058-59; McCollister, Tr. 11041-43; Welke, Tr. 17380-81,
la "
17
i
* Maintenance was included in the lease prices for equipment
that the manufacturer continued to own and the user leased.
19 ;| Maintenance was priced separately for purchase customers. (See
;| Spangle, Tr. 5094-97 (Honeywell); Macdonald, Tr. 6980 (Burroughs);
20] Weil, Tr. 7087-88, 7099-100; R. Bloch, Tr. 7804 (GE) ; McCollister, Tr
i 11461, 11476-77 (RCA).) Manufacturers have strong incentives to
21 1 provide such maintenance in order to protect their property. (See
j Norris, Tr. 6069-70, McCollister, Tr. 11476-77; Vaughan, Tr. 21732-35
22 I The significance to the users in the early days of maintenance
1 being included to lease customers, however, was similar to that
23 i of the bundling of other support services — it increased users'
[willingness to experiment with this new equipment and helped to
24 1 assuage their fears and minimize their risks. (See Welke, Tr.
"19225-28; see pp. 53-67 above.)
?«
-451-
17343-46.)*
As a consequence, virtually " [a] 11 the computer manufac-
turers marketed on a bundled basis" during the 1950s from the Univac
I on. (Goetz, Tr. 17500-01; Spangle, Tr. 5092; R. Bloch, Tr. 7604;
McCollister, Tr. 11042-43; see also Norris, Tr. 6066.)
At IBM, the provision of bundled support began before the
installation or even the acquisition of a computer by the customer.
Such support was viewed both inside and outside IBM as an essential
part of the marketing effort. The IBM systems engineer (SE) was
"part of the marketing team" (Akers, Tr. 96554-56) and would
j| assist in the preparation of the proposal made to the customer.
( Id . ; Enfield,** Tr. 19908.) The IBM salesmen drew on them for tech-
!j nical support. It was the systems engineers who "had the implied
; | responsibility of . . . developing systems to make sure that the
| \ * In formulating the provisions of the 19 56 Consent Decree
:j concerning IBM's obligations to customers that purchased equipment
r ! | from IBM, the Department of Justice apparently recognized the benefit
•1 to users of the support provided by IBM without separate charge. The
^ .! January 25, 1956 Final Judgment obligated IBM "to offer to render,
;| without separate charge, to purchasers from it of tabulating or
• '! electronic data processing machines the same type of services, other
"j than maintenance and repair services , which it renders without
i ■' separate charge to lessees of the same types of machines". ( U.S. v
Vj IBM, [1956] CCH Trade Cases § 68,245, Part VI, §(a), (S.D.N.Y. 1956) .:
r !
• I ** At the time of his testimony in 1976, Enfield was President
? * of The Computer Software Company. (Tr. 19841.) Between 1964 and
• j 1969 Enfield was employed by IBM, first as a systems engineer and
. j then as a Product Administrator in Data Processing Division head-
3 i auarters. (Tr. 19843-44.)
-452-
machine was put to good use". (Welke,* Tr. 1700 9.) They worked
with customers to define requirements and in system design, develop-
ing approaches to problems, also engaging in customer education and
training and in programming. Such work would sometimes continue
after installation. (Welke, Tr. 17007-10, 17069-70, 17372-73.)
In short, the systems engineers were responsible for "making sure
that the customer was indeed implementing the targeted applications,
the 'business applications, and doing the job properly and being of what-
ever assistance we could to make sure that the machine was . . .
performing properly". (Welke, Tr. 17010? Akers, Tr. 96555-56.)
Other firms in the industry also provided those types of
services as part of their marketing efforts. McCollister testified
that it was "normal for some fraction of the time of the [RCA] market-
1S
17
IS;:
■*■- j * At the time of his testimony in 1976, Lawrence Welke was President
,j of International Computer Programs, a firm providing "an information
2G 4 service to the computer software product marketplace" by publishing
i catalogs of software products and by conducting seminars on buying
— •jand selling software. (Welke, Tr. 17003-04.) Welke's first job in
•j 1954 was with General Electric and he had the responsibility of
22 j installing a punch-card system in GE's production department. Between
j 1956 and 1963 Welke worked at IBM as a systems engineer for three
23 j years and as a salesman in the Data Processing Division for four
jyears. Between 1963 and 1968 Welke was with a consulting firm and a
2- ) bank as head of their automated customer services division. (Welke,
JTr. 17004-07.)
25 j
i -453-
"salespeople, maintenance people and systems analysts and programmers,
technical people" as "a normal and as a necessary part of the success-
ful sale and installation of computer equipment." (Tr. 11370-72.)
I' In 1972 Ray Macdonald, President of Burroughs, stated that:
"A major element of the marketing effort in our industry
is support activities. It is important to note here that our
industry's involvement with its products lasts throughout the
lives of those products. It starts with semi-finished raw
materials, continues through intermediate and final manufac-
turing processes, and extends to a full range of services in
support of the product throughout its use.
"At Burroughs, we developed a worldwide capability in
excellent technical support of our products very early in
the traditional product period. With the introduction of
the computer, we have significantly extended our support
operations by adding the new dimension of supporting the
I customer in his use of the product. This includes systems
y j planning and installation support, and perhaps most important
of all, the support of the customer in his application soft-
ware requirements." (DX 426, p. 12; see DX 427, p. 4.)
I
c
a
7 |
Si,
»1
: account varied and not in any simple way. As Akers testified,
:j systems engineers at IBM "were a scarce resource within the branch
4 office" so there was an attempt "to manage the technical talent in
J a way that was most beneficial in [IBM's] sales efforts and installa-
2 i
| tion efforts with our customers." Systems engineers "were allocated
J- "i
ion the basis of how much assistance a particular customer needed at a
Similarly, in 1961 NCR reported to its stockholders that its "marketing
organization . . . provides necessary programming aids, training
courses for the customer's employees, technical assistance on site
preparation, and other supporting services of various kinds." (DX
402, p. 10.)
Obviously, the amount of SE services needed at a particular
particular time; the degree of experience that the customer had;
; -454-
i
i
L
2
3
4-
3
y
3
9- i
ia
xi
iz
13
. /*
17
18
IS '■[
* i
i
whether or not that customer perhaps required additional educational
effort because he or she was installing a new computer system or
computer system for the first time. It was an effort to try to use
that resource as productively as possible in pursuing the quota
objectives that the branch office had." (Akers, Tr. 96555-57; see
Enfield, Tr. 19878-79, 19886-88; DX 4793.) Systems engineers were
assigned to customers on the basis of "who needed the work done and
what had to be done to make it a successful installation". (Welke,
Tr. 17017.)
Systems engineering services were provided to familiarize
users with computers and to ensure that the user, if he chose to
acquire a computer, used it properly to solve his problems. Such
service relieved users from some of the risk of acquiring a computer
in order to induce them to acquire it in the first place. But, in
relieving customers of such risks, IBM, like other manufacturers,
assumed them. By giving the users "a predictable cost that they
could budget against" (Welke, Tr. 19225-26) , the manufacturer took
over the uncertainty in cost resulting from unforeseen variation in
user needs.
Concomitantly, manufacturers stood to gain (by lower costs)
( if over time the customer required less or no assistance. In the
J long run, the reduction in customer needs would be accomplished in
part, as it turned out, by the provision of increasingly sophisticated
operating systems relieving customer programmers of a number of
24. 1
j complex tasks, but it could also be accomplished in the short run by
-455-
*
T
\
}
}
L
? !'
'I
2
training customer personnel in the tasks which software had not yet
taken over and in the use of the software- hardware combinations.
Thus, according to the IBM "guidelines" concerning programming,
"Systems Engineering personnel were to clearly encourage self-suffi-
ciency among the customer [sic] in his programming capabilities with
regard to application programs." (Enfield, Tr. 19862; see also
Welke, Tr. 17373-75.) Such self-sufficiency "was a self-serving
objective. The objective to enable the user to provide more of his
own support would enable an SE to perform less of those functions",
freeing the systems engineer for other assignments. It would also
make the customer more efficient if he did not have to depend on
others. (Enfield, Tr* 20249-50.)
In November 1962 Prank T. Cary, at the time Vice President of
Field Operations for the Data Processing Division, put out guidelines to
IBM executives, regional and local management and sales representa-
tives and systems engineers saying that it was IBM's "responsibility"
to provide to its ■ customers "the assistance they need to install and
obtain the results from the use of our equipment that we have outlined
'{in our proposals to them". Among IBM's responsibilities were the
9 I
'\ "[e]ducation of customer personnel" and the provision of "[tlechnical
5 i
j guidance" in "the use" of IBM equipment and in "programming and test-
II !
jing". Similarly, in order to underscore to the recipients the extent i
2. ! !
! to which IBM was committed to having the customer assume responsibility, •
'•* I i
'the guidelines emphasized that it was the "customer's responsibility" !
14 : I
| to "[w]rite his own operating programs", " [w] ire the necessary control
?« i !
I ;
! i
\ -456-
I
2
3
4-i
f
7
3
panels", " [o]perate the equipment" and "[p]rovide for the physical
installation of the equipment". (DX 4793.) It is worth noting
that these guidelines applied only to lease customers and to "the
first user of purchased equipment" to whom IBM felt it to be its
"responsibility to provide . . . the assistance they need to install
and obtain the results from the use of our equipment that we have
outlined in our proposals to them". ( Id. ) IBM, quite naturally, was
making this marketing support available to its customers and not to
5 ; users that acquired IBM equipment from other sources.
10
12
13
Ralph A* Pfeiffer, Jr.,* described IBM's philosophy as follows
"What we were trying to do was to insure customer's
profitable use of the equipment. The Manager has a certain
stable of talents; he had a customer set that he had to
support and he tried to make the most productive, efficient
use of that cadre of personnel.
"We are trying to supply a service to a customer. We
are trying to have that customer make profitable use of his
equipment. And if he is unable for some lack of whatever it
might be, education in a certain area or a certain person who
9/m . ' he relied on left and he was caught short, we try to supply
10 ij that missing ingredient until he is able to handle it himself.
■ We tried to train him.
"We certainly were interested in having him be capable
~-S |j of running his own installation in a profitable way. Whatever
:j that required in the way of training somebody or supplying that
1$ :{ piece of education that was missing, I hope I operated accord-
ingly." (Tr. 16019-20.)
i
21 \
4
22 i
t
.- i
The policy of building self-sufficiency in customers, however,
j *Mr. Pfeiffer is an IBM Senior Vice President and Chairman of the
•; Board of IBM World Trade Americas/Far East Corporation. (DX 30 74, pp.
2- 142-43.) At the time of his testimony he was an IBM Vice President and
I the President of the Data Processing Division. (Tr. 2963-64.)
^5 !
-457-
carried with it an end to the practice of not charging separately for
such services. When enough customers became self-sufficient and when
changes in hardware and software ceased to require them to be taught
i.J very new ways of operating/ i't would no longer make sense to bundle.
By increasing self-sufficiency in customers, IBM created a growing
group of customers who did not require the bundle. The exact date
r j| on which that group was sufficiently large that it made sense to un-
bundle and provide the formerly bundled services at separate charges
for those who wanted them is a matter of judgment. As we shall see,
in IBM's judgment it came in 1969.
b. The Continued Demand For Bundling in the 1960s . During
the 1960s, for the same reasons as in the 1950s, most users continued
to prefer the bundled offering. (We Ike, Tr. 19230; R. Bloch, Tr.
7751-54.) The demand for such support was not restricted to new
customers unfamiliar with computers. Even in selling to "the large,
I established user", such services would be required "to some degree".
;i "[T]here would always be areas which are unfamiliar to even a relatively
j sophisticated customer. The fact that he was graduating from some
i smaller system to say, a larger . . . system which might involve
i
| communications, this communications area would be the first time for
,i
j that large customer. ... So even with sophisticated customers these
1 ••
j kinds of support were required." (Beard, Tr. 9944-46.)
>
1
I
1 i
I
L
e?
■*
s
7
S
S
.0
_£. »
] The demand for support services continued in the 1960s as
users were rapidly exploring new computer uses and as software
improvements and architecture changes were occurring at a rapid rate.
La
M
-458-
Li
2 ;
1 1
4*:
5
e
7
3
?
10
II
12
13
14
iO <
As Beard testified:
"... most of the customers we [RCA] were dealing with
in the time frame of 1960 to 1970 -were not thoroughly experi-
enced in the use of data processing equipment. The field had
gone through a very dynamic growth. It faced new technology,
a new set of programs imposed upon the business organizations
that used computers. So a lot of people felt they were on
very shaky grounds. They were not sure of themselves."
(Tr. 8497.)
New products and new ways of doing things were being introduced requir-
ing customer training, programming and systems design services,
imposing additional demands on the manufacturers. (McCollister,
Tr. 9647-53; DX 69, pp. 3, 5? see also DX 98, p. 12.) As Withington
observed in 1968:
"Programmers and system analysts are in inadequate
supply. This shortage has existed for years and shipments
have nevertheless grown, but in one major respect the problem
is worsening. The advanced, integrated applications many
users wish to implement are novel and very complex and require
much more creative, high-level system analysis than the
simpler, second generation applications did. Since experi-
enced system analysts are in the shortest supply of all, this
pressure may have an increasing effect.
"The increasing complexity of the third generation
'] hardware and' software (a necessary corollary to its increased
i ' |l capability) makes it difficult for the average user to under-
I stand and use. It may take longer than it used to for users
^ i to fully exploit the equipment they are currently installing:
;j many users will not be able to use anything larger or more
•^ j complex for a number of years." (PX 4833, p. 9.)
20 j Thus, although users would eventually, become familiar with
i
21 | the architecture of System/360, the sharp increase in complexity
j.
22 "\ as users moved from second generation equipment to System/ 3 60 tended
23 j to offset the gains from previous experience. Users were being
2- 'i trained and retrained to use more complex equipment in increasingly
25 j sophisticated ways and the bundled IBM offerings were all the more
t
■i
.j -459-
important to the System/360 user* (See Welke, Tr. 19617-18.) For
example, Welke testified that System/360 "represented a new level of
hardware technology . . * it represented a new level of software
technology with its systems software environment and the very way
»
that you approach programming and processing. It caused a complete
change in how people approached the task of data processing."*
( Id. ) In January 1964, F. P. Brooks, Jr., wrote that:
"The breadth of System/360 and the number of innovations,
particularly in gross systems concept, will require
substantial lead time between announcement and proper
installation. . . . The sheer amount of new abilities, new
options, new specifications, and new prices will require
time for the customer to assimilate. A major education
program for IBM field personnel and customers must intervene
between announcement and successful installation." (DX 1172,
pp. 1-2; see also Withington, Tr. 56591-93; DX 4815.)
As we have seen in the discussion of System/360, all of
this happened — and more — with the result that IBM was compelled to
expend tremendous effort and expense to install and support System/360.
As other third generation equipment began to appear, other manufacturers
found requirements for support services growing as well. SDS told its_
stockholders in its 1965 Annual Report that
r
i
U
2
3
7
3 •!
; | "Ct]he character of the computer market changed substantially
a j last year as the result of advances in both the understanding
•j of the technology and in the manner in which computers should
a !
i
T i
2 ;| * "By [1968] the marketplace had acclimated to these new hardware
1 technologies and software technologies. By the same token, the
3 j product, the 360 system, particularly with reference to the software
j involved, the system software, had settled down and achieved a
m 'j respectable semblance of predictability." (Welke, Tr. 19617-18.)
-* i
-460-
L
2
Z
4-
S
7 i
3
9
10
II
12.
13
14-
15
17
IS
IS
be employed. . . . During the past year increasing emphasis
has been placed by management on providing complete service
to SDS customers both before and after installation. To
this end, technical staffs and applications programming, systems
engineering, customer training and maintenance have more than
doubled in size and in the scope of their activities*" (DX 981,
pp. 4-5.)
Such increases continued for the next few years. (DX 982, p. 3; DX 983,
pp. 16-17.)
Similarly, RCA found that the introduction and installation
of its new Spectra series created large user demands for assistance.
(McCollister, Tr. 9649-53, 11403-06.) NCR told its stockholders in
early 1966:
"As the trend toward fully integrated business systems gains
momentum, NCR's opportunities for growth and greater profit-
ability can be expected to increase proportionately. Full
realization of these opportunities will require an aggressive
continuation of the program of recent years. To this end,
additional expenditures will be required not only for further
product development efforts but also for training sales and
service personnel and for providing the many supporting
services essential to the successful marketing of advanced
business systems." (DX 368, p. 3.)
It was a view which NCR was to reiterate as time went on. The
following year, it stated:
"Today, a . . . requirement for future success in the
marketplace has arisen; that is the need for business equipment
suppliers to provide additional guidance to customers in the
utilization of new technologies for operating their businesses
20 | more profitably. For in the final analysis, the effectiveness
■j of today's sophisticated information systems depends upon a
21 ;J - full understanding of their potential at all levels of manage-
j ment. To this end, NCR's educational programs are being
22 1 designed not only to prepare sales representatives to install
] advanced systems, but also to provide counsel and training in
22 ; management sciences." (DX 370, p. 5; see also p. 19.)
i
24. Sin its 1967 report, just after the announcement on March 2, 1968,
25 I of its new Century series, NCR reiterated this position,
i
!
:! -461-
3
1
T
z
m
3
4»
"In addition to offering outstanding equipment, meeting
the demands of the market today also requires:
"1. Expert diagnosis of customers' current and future
business information requirements , based on broad systems
knowledge and experience.
"2. A complete range of supporting software, including
standard programs for many applications and in-depth training
of customer personnel.
"3. Continuing support of every installation, with
upgrading of both system and equipment as customer requirements
change .
"The company's marketing strategy is based on providing
this full spectrum of customer services." (DX 366, p. 3.)
Similarly, one of Sperry Rand's major objectives in 1962 and 1963
was to "give increasing emphasis to our computer service and
marketing". (DX 69, pp. 3, 5.) It was an "emphasis" which R. E.
McDonald was to look back on in 1973 as "[o]ne of the main factors"
behind Univac's success. (DX 98, p. 12; DX 65, p. 2.)
c. IBM's Unbundling Announcement. On December 6, 1968, IBM
;| announced that it expected "to make changes in the way it charges
;) for and supports its data processing equipment" during the following
7 il
;| year. (PX 3390.) It announced its decision in detail on June 23,
8 3
: 1969, with the changes effective immediately for new orders and
I effective January 1, 1970, for customers with machines installed or
I on order. (PX 3351; PX 3352.) Basically, the announcement instituted
- !
•ij charges for systems engineering services and education and for new
•2 )
■ "program products, as distinct from system control programming". (PX
3351, p. 4.) Programs then available from IBM's library continued
l± \
.{to be available as in the past at no separate charge. I3M also
IS ' ]
]
\ -462-
L
2
3
4-
5
7
a
9
10
11 ]
12
13
14* ;
15
16
offered to engage in contracts assuming "responsibility for the
performance of specified tasks in the areas of systems design and
analysis, application and program development and systems install-
ation and evaluation" . ( Id- ) No change was made in the way in which
maintenance was provided, maintenance on purchased equipment continuing
to be available at a fee and maintenance on IBM-owned equipment
leased to users available without a separately stated charge. (PX
3351; PX 3352.) IBM also reduced its prices by 3%, stating that
this reflected its "best approximation" of the expenses which would
"no longer be provided for in prices of currently announced equipment".
(PX 3351, pp. 1-2.)
There were a number of reasons for the announcement.
First, IBM, like others, was feeling the strain of standing ready to
supply services on demand without an extra charge in an increasingly
complex environment. IBM "stated that — as a result of fast-changing
i data processing market conditions — the need for increasingly complex
and comprehensive systems support is growing more rapidly than antici- "
I pated. In addition, new support requirements are arising from leasing
;j companies and other owners of IBM equipment as they relocate and
^reapply their systems." Such demands for "new and additional forms of
i support services" were expected to continue to grow. (PX 3390, pp. 1-2;
21 ! .
IPX 3351, p. 3.) As would be expected in a company accepting the risks
I and burdens bundling entails, there was a recognition within IBM of
22 j
[the increasing costs of providing software and support. During the
24 i
25
j early and mid 1960s persons within IBM observed that programming
-463-
expenditures were "skyrocketing" and "increasing dramatically" (PX
2804A, pp. 1, 2; PX 2805A, p. 1; PX 4053, p. 1), and attempts were
made in 1966 to quantify the return to IBM on programming expend!-
' tures. (PX 1748, p. 11.) Cary testified that increasing demands of
customers for education led to separate pricing of certain education
offerings and that IBM was "always looking for ways of reducing the
cost of systems engineering". (JX 57, p. 2.)
The general problem of cost escalation was magnified by
the special problems associated with installation of System/360. As
we have noted earlier, because of the unprecedented — and unantici-
pated — success of System/360, IBM had added new people to its marketing
division. (See p. 372 above.) The training of such people, the
support required by users to effect their conversion to the new and
sophisticated operating system software associated with System/360, and
the problems which IBM encountered with some of the 360 software caused
IBM to devote an enormous portion of its resources to supporting the
j installation of System/360 and making sure that customers were able
it to do their work during the transition phase. (See pp. 369-72 above.)
i
j The result of this, however, was that levels of support far greater
5 :|
| than ever before required were demanded of IBM. The cost of providing
m I
i such support had to be borne directly by IBM itself, but in the long
■J run, of course, it would have to be absorbed by IBM's data processing
2 ; i
| users .
2 1
! At the same time, by 1969, in part as a result of I3M's
4 1
i policy of encouraging self-sufficiency, there had developed a group of
el"
-464-
|
z<
si
i
5:1
I
7;
3 I
9 I
i
m
n
12
a
15
* *•-
17
IS
13
relatively efficient and sophisticated users who would accept much
more of the risks of computers and Were willing to do much of the
support in-house* As Welke testified:
""The more sophisticated users, and the ones who had the
best-run or the best-managed shops, for the most part were
ready to accept the idea of unbundling, because I think they
saw in it a chance to be more cost effective in their entire
data processing operation." (Tr. 17172-73.)
T. J. Watson, Jr., testified:
"We had some very sophisticated customers by this time,
Lockheed, Boeing and others, who felt that they were better at
performing, some of these services than we were. They felt it
onerous to pay for them when they, themselves, could do it in
their opinion better." (Tr. 16602.)
Another reason for IBM's announcing unbundling in 1969 was
that, by that time, the notion of charging for software and services
had become relatively accepted because of the entry and success of
software houses. (See below, pp. 851-65. ) That had not always been
true, however. From the early days of the computer industry up until
the late 1960s software was generally looked upon as something other
than property that could be appropriately charged for. "For the
longest time, computer programs were looked upon as an intellectual
product, but not necessarily having proprietary value." (Welke, Tr.
: ( 17361-62; see also DX 1096, pp. 1-2.) This led many people to believe
20 :
i that in fact most users were not willing to accept the notion of
21 :| .
i software as a "product" in the 1960s. (See Welke, Tr. 17093-95, 19180-
J 82.) This view was both illustrated and reinforced by the free
23 =|
interchange of software that was characteristic of this period. (See
24 "■»
i below, pp. 856-58.) But, during the 1960s software houses began to
25 i
I charge for software products that competed with IBM's unpriced offerings
-465-
and by 1969 " [t]he industry had developed to a point where many of those
services were available, separately, and outside." (Watson, Tr.
16601; PX 3351, p.. 3; see pp. 858-59- below. ) .* .As a result, IBM began to
if believe that, for the first time., there might be business opportunities
in selling software and services separately. (E.g., PX 3351 r p. 3.)
Under such circumstances, it was possible for IBM to stop offering
such services, which it was finding "onerous", under the bundled system.
IBM's Chairman testified that it "seemed like an appropriate time,
from a business standpoint of view, to open the matter up in the way
that we did". (Watson, Tr» 16602.)
Not surprisingly, customer reaction to IBM's unbundling
r I announcement varied. Some relatively more sophisticated customers
welcomed unbundling; others, generally the relatively unsophisticated,
were less happy. Welke testified:
:l
t '
3
r I
i
"The initial reaction was — it varied. Some people were
very happy with it. A good number of them that I came in con-
tact with were anything from hostile to total disbelief as well
"I think it depended on the sophistication of the user.
The more sophisticated users, and the ones who had the best-
;j run or the best-managed shops, for the most part were ready to
3 ;l accept the idea of unbundling, because I think they saw in it
|j a chance to be more cost effective in their entire data process-
' : ; ing operation.
T
j
,
.| * One of the reasons for this was that industry practices had emerged
3 •} which gave sellers some assurance of protection of proprietary program- j
j ming from plagiarism. (Welke, Tr. 19211-13.) Welke testified that j
A I "many sellers" of software at one point, and "to a very limited degree" j
;! still, feared that their software would be plagiarized because of the j
S j ease with which programming can be copied and the inadequate protection !
i
t ;
I
i ;
:! -466-
z
3
4-
3
«•
a
7
3
?
10.
II
VL
13
14
15
^ ,
17 \
IS :
aa :!
"The ones who didn't want the unbundling or who were
against the idea I think in some cases were also the ones that
were getting, . . . more than their normal share of IBM's
support and systems engineering and programming as well . "
(Tr. 17172-73.)
. Similarly:
" [Ulsers, even in 19-69, when they heard about unbundling,
were reluctant to accept it or were hesitant and in some cases
even hostile to the idea. At that point in time users were
beginning to get a pretty good idea of what some of their cost
elements were and the more sophisticated, more advanced
users had a way of breaking out cost, elements in thei'- total
computer operation, identifying them, and controlling them.
"But for a lot of users, there were still many, many
unknowns in their data processing operation, things that they
didn't know could happen, they had no way of anticipating, and
I think they wanted the assurance that bundling, in effect,
offered them, that one way or another, if and when the unknown
occurred, they'd be covered. It was an insurance policy in
many respects." (Welke, Tr. 19226.)
As might be expected, reactions of other manufacturers varied
also. Bundling had been a practice desired by users. Users' needs
changed over time as they became more sophisticated and self-suffi-
cient but this was a continuous rather than a discrete process , and
opinions, even in. 1969, could very well differ as to whether the time
had come to make the changeover. For many companies the decision
;j whether or not to unbundle was not entirely a foregone conclusion.
22, j afforded software through patenting and copyrighting. (Tr. 19211-13
!j This was recognized within IBM during the mid-1960s. In 1965 it was
22. | stated within IBM that "an overriding factor against unbundling
: j [certain programming] is our present inability to protect the pro-
23 1 prietary use of our programming systems". (PX 1651, p. 6.) In 1965
I R. H. Bullrsn, then an IBM Vice President and Group Executive, wrote
24 i that: "We must settle on whether or not, and to what degree t
\ we can protect programs before we can deal adequately with
25 I the question of selling them". (DX 1031.)
I
!
t -467-
1
2
3
r
6
7
8
9
10
ii
12
i
13
14
15
16
17
18
19
20
21
22
Spangle, of Honeywell testified that Honeywell did not
follow IBM's lead for a number of reasons. It was not set up adminis-
tratively to charge for the separate items and enforce their collection
throughout the field; it was not certain of the contractual arrange-
ments it had. with its existing customers; finally, Honeywell "hoped to.
gain some temporary market advantage . . . because we thought there
would be quite a bit of resistance to this change by the customers and
prospects, and that because of that we might be able to get some
customers that we otherwise would not have been able to get". (Tr.
5086-87; see also Withington, Tr. 56786-87.)
Univac had similar reasons for not unbundling when IBM did.
McDonald testified:
"Actually, we felt that there would be considerable
anxiety in the marketplace as the result of IBM's decision and
announcement to unbundle, and we felt it would be to our
competitive advantage to maintain our previous pricing policy
so that we could go to the customers, potential customers of
IBM, and say to them that we would offer you these services
which we have in the past under the same pricing policy, and
you know what you will be getting from us, and under the IBM
unbundled pricing policy, only time will tell what your real
prices will be; and I think this was effective, at least for a
period of time*
" [W] e did see some increase in bookings over what we
expected our bookings would have been had IBM not changed their
policy . . . which we attributed to IBM's unbundling." (Tr.
2896-97.)
Similarly, McCollister of RCA "recommended that RCA
I; should continue in the business by continuing to offer bundled
24 || services". He felt this to be "to the benefit of RCA in its
relationship with its users .... [G] iving assistance to the user
as required could lead to and usually did to a more effective use
-468-
25 :
Li
2
3-
of that equipment by the user and gave RCA, therefore, a stronger
installation and to the extent that the equipment was on rent,
insured more completely a continuation of the rental income." It
4»l r brought in more money than unbundling. Further, "I felt, this had been
a sound business policy for the IBM. company for a long time and just
because IBM . . . decided that they would change, I did not see at
7 1. that time that this was a reason for the RCA company to change and do
3
2 :
ia
11 :
12
13
1^:
15
IS !
17
l*|
2a j
■1
TT :|
22 1
differently" . "I believe the customers preferred the method which RCA
had been following and which RCA elected to continue." (Tr. 11206-09.)
NCR went some distance in the direction of unbundling.
On October 1, 1969, it stated its belief "that each user of its
computer systems must be provided with a certain essential amount
of software, systems support, and educational services if he is to
successfully install the system and begin to benefit from his
investment. NCR believes that this basic package of supporting
services must be the responsibility of the equipment manufacturer."
(DX 346, p. 1, emphasis in original.)* NCR recognized that there
* NCR expressed the view in its 1969 report to stockholders that,
"The deluge of new concepts and new equipment which has
flooded the information processing industry in recent years
points up dramatically the need for ever-greater customer support.
•" j "Indeed, the growth of the industry will continue to depend
\ in large measure upon its ability, through supporting services,
2- I to adapt these new concepts and equipment to the requirements
.1 of different organizations. Thus, during the 1970s increasing
25 I funds and effort will be devoted toward broadening the spectrum
i of customer assistance." (DX 367, pp. 19-20.)
-469-
J
J 1
would be considerable variance in the level of support required by
different customers and stated that "it will continue to be NCR's
policy to provide, as part of the basic hardware price, that amount
^ | of software and support which will realistically insure that a
prudent user will be able to install and successfully utilize his
NCR computer system". An allowance, based on the size of the
system amounting to "approximately 30 man-days of support for each
$1,000 of monthly rental" was to be provided with support above
that level billed separately. The same principle was to apply to
educational support and software "including both applied programs
and computer languages". (Hangen, Tr. 10721-24; DX 346.)
2 | On January 1, 1970, however, NCR announced a change in
3
c
its policy stating:
"After further evaluation, it has been decided not to
price all basic and applied software and not to establish an
allowance against which such chargeable software would be
applied. The NCR software pricing plan will be to continue to
establish pricing for software products on a selective basis,
considering the value to the customer, uniqueness, and other
factors." (DX 386, p. 2, emphasis in original.)
•t
■I
si
rr 1
i
>2 1
There was much less disagreement in 1969 and 1970 on the
question of whether or not operating systems or systems control
i programming should be unbundled.* IBM did not unbundle such pro-
gramming, stating: "System control programming is an essential part
" 1 * By the early 1970s only CDC had unbundled its operating system.
1 (Norris, Tr. 5647; Goetz, Tr. 17530.)
1*. \
-470-
L
Z
3
4*
a
T
3 ;
i
i
ia i
t
I
U !
12
T
of a data processing system. It is fundamental to the operation and
maintenance of the system and will be made available as part of the
system," (PX 3351, p. 4; see also PX 2454, p. 1; PX 3352, p. 5.)
The fact that operating systems were essential was widely
v
recognized, Enfield testified that he did not "see how" a supplier of
computer systems could market its equipment without making available
some form of an operating system, at least following the introduction
of System/360, by either producing the operating system itself or
arranging for it from some external source. (Tr. 20740-41, 21074.)
Welke testified to the same effect, (Tr. 19223.) Dr. Perils of Yale
testified that
LZ
"operating systems are . . . indigenous to all major computers
at the present time. They manage the computer resources and
they really could be part of the hardware except that their
functions are not well enough understood at the present time
ti • to make it economically feasible to put it into hardware".
(Tr. 1344.)
IS
According to Perlis, operating systems are "crucial to the
successful operation of almost every computer around today".
u ■ (Tr. 1348.) As a result, operating systems are "typically" designed
to ;
•* it for a particular fit with a particular computer, because "[t]hey
i9 | depend very strongly on the particular resources and the way those
; i
& j resources are organized in a particular machine, and they do this
:l
~ ; so that they can achieve the most efficient operation possible . . .
22 :j to take maximum advantage of the idiosyncrasies of the hardware".
** j (Perlis, Tr. 1986.) Thus, Withington testified that hardware and
2- I software "are now necessarily designed as one, designed to execute
25 j f rom the same architecture". (Tr. 55919-20; see also DX 491, p. 5.)
1
I -471-
3
Hence, separate pricing of operating systems would require "arbitrary
allocations".. (Withington, Tr. 56798.) Indeed, Withington wrote in
June 1969, shortly before IBM's June 23 unbundling announcement, that
■ systems software was "essential to the operation of modern computers
and is designed contemporaneously with the machines. It is not
;j possible to separate its development costs from those of the computers
themselves, nor is it possible for the machine to operate without some
version of the operating system" . He concluded that this was a
"complex area" and that "basic skeletons of the operating systems"
were, at that time, "likely to be provided free with every machine
. . . because there is no rational way to separate them". (PX 4834,
p. 11.)
Ray Macdonald of Burroughs testified:
"[W]e had extensive discussions on systems software, and I
believe that our conclusion after some experimentation, and
quite a bit of back and forth debate, was that the systems
software that I have described is in fact an inseparable
part of the system for the average user.
m ' "Now there may be the very unusual user which represents
r -j an extremely small portion of the total market who may . . .
have the sophistication to consider a different mix or differ-
5 . ent system software for his own purposes, but first of all, I
j think this is a very, very small portion of the total market,
* : j and certainly not suitable for the vast majority of the market."
; (Tr. 6977-78.)
3
i
T ;l
Thus, as the 1960s ended IBM had embarked on a course of
4 separately pricing certain of its software and services . As we shall
| see, IBM continued on this course during the 1970s, separately pricing
| increasing amounts of its software and services in response to rapidly
4 i
! changing market requirements and technological advance.
! -472-
■l _. . - , , „ „ i
14- ;
. - it
1/ •
(DX 10; Eckert, Tr. 1014-17), Univac had manufactured and was still
marketing several incompatible product lines (represented in 1964 by
the 490, the Univac III and the 1107) , each requiring different soft-
j ware. Moreover, Univac had failed to provide successors to its
obsolete products. (See PX 4829, p. 20; DX 8, pp. 1-2; DX 10; DX 14,
' In 1964, "after it had become apparent to the rest of the
1S:|
..p. 1.)
19 i
za :|
1 industry that magnetic disks were superior" , Sperry was still marketing
.j its FASTRAND drum instead of quickly proceeding with disk development,
22 'I
■I * R. E. McDonald was President of the Univac Division from 1966
24,; to 1971. (McDonald, Tr. 2769, 2776-78.)
ys i ** J. P. Eckert was a Vice President of the Univac Division and
'technical advisor to the President of Sperry Rand at the time of his
:! testimony, having held that position from "about 1960". (Eckert,
; {Tr. 710.)
'!
i| -473-
40. Sperry Rand/Univac. Although Sperry Rand's Univac Divi- j
sion entered the second half of the 1960s lagging substantially behind
the industry leaders in the areas of product compatibility and storage
technology, it was able, by the end of the decade, to reestablish
itself as a major' force in EDP, logging substantial gains in revenues,
organization and technology.
a. Univac* s Problems in 1964. Univac, in 1964, was in a
state of some disarray * It was in the midst of a succession of
presidents (Eckert, Tr. 1008-13r McDonald, Tr. 3785-88)* and was "still
suffering" from the "great drawback" of its "inability to assemble
a smoothly working, reasonably permanent management team". (PX 4829,
p. 20.) Additionally, despite the suggestion of Dr. Eckert**
i
5
3
7\
3
2
a
z
.6
J
,S
19
za
a delay which had a substantial adverse effect on the marketing of its
computer systems. (Withington, Tr. 56455, 56485-87.) Consequently,
Univac was compelled to purchase disks from other suppliers " for a
while".* (Withington-, Tr. 56243-44.)
Univac'.s financial results during the first half of the
decade were not particularly encouraging. In 1962, the corporation
had found that "the rate of technological obsolescence" required it
. to write down the value of its older EDP equipment by more than $50
million, and to accelerate the depreciation of its newer models.
(DX 69, p. 3.) In 1964, Univac was "losing money" (McDonald, Tr. 3813)
and experiencing a relatively slow rate of revenue growth. (14.8%
compound growth rate from 1960-64 compared to 27.4% from 1956 to
1960.) (See DX 8224, p. 624.)
Notwithstanding its limited success , Sperry reported to its
shareholders in its 1965 Annual Report:
"Data processing is a dynamic industry, having great growth
potential. It has established a place in the world's economy
j that is essential and will continue to grow. Such dependence
! upon any industry in the past has not only led to growth but
also profitability. Therefore, we have determined that we will
! remain in and grow with the data processing business." (DX 13983,
I P. 6.)
i * IBM employees reported, in 1968, that suppliers of disk drives
21 i to Univac included Vermont Research, Bryant, Data Disc and Memorex.
: i (PX 2267B, p» 27.) Univac continued to purchase disk drives from
22 :j Memorex through 1970 (Guzy, Tr. 33170-71), and from CalComp through I
| 1973. (PX 5584, p. 16.) It also purchased disk drives from Peripheral !
22 | Systems Corp. in 1969. (DX 1302, p. 1.) !
i
1 -474-
L i
2 i Univac concentrated its marketing efforts on the Federal government and
5 !
I
5 i
7;
i
3 j
i
i
IQ !
!
i
II :
i
IZj
13 ;
According to McDonald, during the period from 196 3 to 1971
airline reservations users. (Tr. 2890-91.) The Federal government was
a very important customer for Univac in. the 1960s, as it was for most
of the industry; Sperry reported that in the fiscal year which ended
March 31, 1964, for example, the Navy had ordered four 490 systems for
world-wide inventory control, the Marine Corps had ordered three Univac
III systems far similar applications and the Air Force had "ordered more
than 150 UNIVAC 1050-11 systems, as well as three UNIVAC 1107' s for
logistic control purposes". (DX 13913, p. 16.) The Air Force order
alone was, as Withington noted, "large enough to cause a bulge in ship-
ment statistics". (PX 4830, p. 22.) During the 1960s Univac claimed
"a complete array of computers" for the military. In its 1964 Annual
Report, Sperry Rand contended that "no other company in the industry
[could] match this range". (DX 13913, p. 12.) Several computer systems
were offered by Sperry Rand to satisfy shipboard, airborne, van-mounted
'land aerospace military and space requirements.* Univac was the prime
'{supplier of the militarized AN/UYK-5 & 7** which were the standard
milspec computers for the U. S. Navy. Univac also had a broad range of
17
13
15 ;
•i
20 ]
'! * Sperry Rand also had large commitments to the space program.
Zl iFor example, eleven Univac 1218 systems were to be delivered in 1964 to
|NASA tracking sites to assist in tracking the Gemini Capsule. A Univac
22 .1218 was selected for the "mobile-wing, limited-warfare intelligence
[complex". (DX 13913, p. 12.)
. j ** The AN/UYK 7, a chird-generation computer, used a general
2± '.purpose software package called Gipsy, developed by the Naval Electro-
nics Laboratory in San Diego. Gipsy provided the capabilities of a
?= master control and data base handling program with a maximum degree
'of hardware independence. (DX 5117, p. 1.)
.1
: i -475-
computers oriented to Navy milspec requirements which were in popular
use aboard. Navy vessels, performing a wide variety of applications.
For example, the Naval Tactical Data System (NTDS) application aboard
• the U. S. S. Enterprise had a Univac USQ20 (1206) as their central
computers. (DX 69, pp. 12-14.) In addition, the AN/UYK-5 computers
were used on Navy vessels to process maintenance records, supply and
accounting applications . (DX 5123, p. 3.) The Marine Corps used the
Sperry Rand 1005 systems (AN/UYK-5 [V] ) for their field van-mounted
applications.. Van-mounted 1005s were also used by the Army in their
PERMACAPS and DLOGS systems in Germany, Vietnam, Korea and around the
United States. (DX 5410, Fullerton, pp. 36-37.) At the White Sands
Missile Range (WSMR) in New Mexico the Sperry Rand 1218* was used for
a variety of applications including: missile guidance and tracking,
data reduction and analysis, simulation, communications, logistics
management, and satellite tracking. (Plaintiff's Admissions, Set II,
H 768. 0-. 4.) The Univac 1108 at WSMR was also used for missile guidance
; j and tracking. ( Id. , 1f 748.2-. 4. )
7 ii
j The Naval Electronics Laboratory in San Diego acquired:
3 1
:| 1 IBM 360/65; 1 CDC 8090; 1 Sperry Rand 1230; 5 Sperry Rand USQ20s
■9 j
I (CP 642A/B) ; and 1 Sperry Rand AN/UYK-7 (CP 890.) (Id., <i 702.0.) As
» ]
i of 1974, applications of a general data processing nature previously
- i •
\ processed on the Univac CP-667, USQ-17 and USQ-20 computers were to
be transferred to the IBM Svstem/360 Model 65 along with those that
23 !
! had been run on the two CDC 1604s. ( Id. , If 702.15.)
14. i
f
L
t
x
r
i
i
j
l ;
Z !
3
4-
I * The militarized 1218 and the commercial UNIVAC 418 are identical
■! in design and the mainframes do not vary at all. (DX 90 8 8.)
-476-
Li
Z
z
3
a
7
3
*!
10
Il f
12
13
L*.
15
Iff
17
IS
19
'I
li
22:1
The Department of Navy reported the use of Sperry Rand
AN/USQ-20s, AN/UYK-5s, Sperry Rand 1219s, and a Sperry Rand AN/USQ-17
for the Navy Management Information System for Education and Training.
(DX 2992, pp. 592-93, 1123.)
In the area of" airline reservations, British European
Airways ordered a 490 in 1964 (DX 13913, p. 13); two years later, in
fiscal 1966, Univac reported that it had been awarded the "biggest com-
mercial computer contract ever awarded", a $39 million contract from
United Airlines, "to design and build a computerized information system
that [would] handle United 's needs through 1975". (DX 61, p. 9.) As
it turned out, Univac was unsuccessful in its bid to meet United' s
requirements, and the effort was "aborted" in 1970, with United
Airlines moving to an IBM system.* (O'Neill, Tr. 76015-17, 76231-32.)
b. The 1108. The United Airlines system was to have been
"based on Univac 's 1108 f s". (O'Neill, Tr. 76231.) This computer,
introduced in 1964 (DX 13983, p. 14), was compatible with the thin-
film 1107, and was intended for Univac' s "large-scale users". (DX
14, p. 1.) Withington viewed the 1108 as "technically impressive",
claiming that its "very fast control memory" marked "the first signi-
ficant appearance of integrated circuits in commercial computers . .
. ." (PX 4829, p. 20.) The 1108-11, a "time-shared version" of the
i * "The reason that United decided to terminate that activity was
23 that they concluded that the system being developed [for] United at
{ Univac would not accommodate their projected volume. They subsequently
24. : { installed IBM 360/65S, and later installed IBM 195s for their
.{ passenger service system." (O'Neill, Tr. 76016-17.)
re
-477-
u
f
T
I
*
2
I
Z\
3
£;
S
5
7
,3
3
1108 was introduced in 1965. (PX 4830, p. 22.) The 1108 was not
delivered in volume until late 1966. (PX 4832, p. 18.) By 1967, the
1108 "accounted, for about half the value of Sperry Rand's shipments".
(PX 4833, p. 17.)
Univac- continued to develop and extend its 1108 system
and related machines through the late 1960s. Univac, in fiscal 1969,
announced the 1106; "a smaller, compatible version of the 1108 system".
(DX 3271, p. 5.) In 1967, Univac entered "the data services field with
a service bureau network Of 1108 's directly connected to small computers
on users 1 premises". (PX 4833, p. 17.)
Univac 1108s were employed in a wide variety of commercial
contexts. The 1968 Sperry Annual Report showed a picture of the
Univac 1108 scheduling trains for the French National Railway.
|| (DX 13914, p. 5.) In addition to United' s reservation system, Fuji
Bank Ltd. , Tokyo, in 1969 inaugurated a nationwide on-line banking
system using an 1108, according to the Sperry Rand 1969 Annual
Report (DX 3271, p. 7) ; and the Sun Oil Company ordered an 1108
system in 1968 for use in processing business and scientific problems
! (DX 13914, p. 16), to name but a few examples. As Sperry management
noted in its 1970 report:
» !
" . . . The Univac large-scale computer systems —
£ j especially the 1100 series— are acknowledged to be the most
"" • versatile processors available. The UNIVAC 1108 and 1106
?2 "i systems, in addition to having unparalleled capability for
j scientific and engineering applications , have gained wide
?3 ; acceptance among commercial/industrial users for business
i
i
2* I
;; -478-
I
2
a
4-
a
m
7 1
3
IQ
II
12
12
14
15
15
17
IS
data processing and communications tasks." (DX 13915, p. 5.)*
In March 1969, Sperry Rand management reported that "the
backlog for Univac 1108 computer systems continues at a high
level. ... It provides the Company with an entree into the
market for management information systems because of the computer's
communications and multiprocessing capabilities in both business
and scientific applications". (DX 3271, p. 5.)
Development of the 1108 was not without its problems,
however,- the 1108 operating system, EXEC- VI II, had "major problems
in its initial stages". (J. Jones, Tr. 79631; PX 4834, p. 25.) These
problems, similar to those encountered by other manufacturers with
complex operating systems during the 1960s (see Perlis , Tr. 2002-03;
Weil, Tr. 7217-19; McCollister, Tr. 9694-97; Rooney, Tr. 12132-36,
; 12349-50; Conrad, Tr. 14088-89; Withington, Tr. 56727-31), came
relatively later for Sperry Rand "because it was not attempting to
offer systems programs as complex and advanced as the other competitors
were". (Withington, Tr. 56736.) During the late 1960s, Univac failed -
to deliver operating systems which completely met their advertised
» :| i
;( * NASA's Marshall Space Flight Center (MSFC) Computation Laboratory ;
20 | utilized 1108s to perform both "scientific data processing**' and j
i "administrative data processing". (Plaintiff's Admissions, Set IV, \
21 -j if- 386.0.) The Slidell Computer Complex at Marshall also used two ]
J1108 systems for rocket stage design work, scientific applications j
22 land "some administrative data processing".. (Plaintiff's Admissions, j
Set IV, flK 390.0, 392.0, 394.1, 401.0.) Five Univac 1108s were j
23 (installed at the White Sands Missile Range (WSMR) , utilized by the I
[Army, Navy, and Air Force: two of the 1108s are employed in real-
24 Itime missile performance computations; two others provide back-up,
j batch processing of test data and remote time-sharing ability; and
25 jthe fifth is used for batch processing of classified' data. (Plain- \
i tiff's Admissions, Set II, "[fl 746.4, 748. 0-. 7.)
1 I
'! -479-
capabilities and, indeed, EXEC- VI I I was delayed at least two or three '
years , not meeting its. advertised capabilities until sometime in the
early 1970s.*" (Perils-, Tr. 2003; Wlthington, Tr. 56737.)
c. The Product Line Task Force. The 1108, though successful,
was not an answer to Univae ' s need for a compatible product line. As
we have seen, it was announced at approximately the same time that
Eckert, in his capacity as head of the Gemini Committee, was calling
for unification of Univae' s dissimilar product lines. In 1965, in
the wake of IBM's System/360 announcement, Frank Forster, Univae' s
President from July 1964 to early 1966 (DX 13983, p. 6r DX 61,
pp. 2-3) , set up a Product Line Task Force to review Univae products
and to help him make decisions about their future. (McDonald, Tr.
3804-05; see also DX 13.)
The task force, in February 1965, reported that it believed
that Univae 's manufacturing costs were higher than those of IBM, and
that :
i
r\
i
?
3 ;
i
z
A
.4-
6 r
• "IBM's heavy investment in product research is beginning
7 f to bear fruit. Its developments in circuits, microprogramming
| techniques, memories, and mass storage suggest that for the
3 :| first time in the short history of the industry, IBM has
•| acquired a definite technological leadership; this, together
,3 ;; with our cost situation, may leave us little to sell.
; (DX 15, p. 2.)
:g: i
i In its next report, issued in March 1965, the task force observed that
SI ! •
■] both Honeywell and RCA had committed themselves to the production of
;j integrated computer families (the Honeywell 200 series and RCA's
5 i
Z4 i
;| * As a result of the delay in developing EXEC-VIII, NASA, for
;e j example, was able to renegotiate its contract with Univae to
J include the grant of free computer time as a "slippage" penalty.
;j (DX 5654, pp. 114-15,-231-32.)
;!
I -480-
Li
Z
3
4-
7
a
3-
ia
u
12
n
14-
15 i
Spectra 70 line) in the "tailwind created by . . . IBM". (DX 16,
p. 2.) The report quoted the editor of Datamation ;
"U&TW&C is the big question mark * . • every month until a
new line is announced weakens their chances of success . . .
and it's not clear they'll offer a complete line at all.
Anything less- could, relegate them to the second division."
(Feb.. r 1*65, p. 8$.) ( Id. )
Nonetheless, the task force was unsure whether Univac should try
to match IBM's 360 or take some other action. Specifically, it
expressed the concern
"that the RCA and Honeywell moves, although based on clever
sales strategies, may not make such good sense financially.
Both are based on the assumption that now that IBM has made
its move, the pace of obsolescence will slow down, and longer
writeoffs will be possible than in the past. It is our
opinion that in about five years this assumption will prove to
be catastrophic to anyone who bases his product line on it."
(Id.)
Ultimately, Univac decided not to introduce a full
spectrum product line but to introduce only three machines , called
models A, B and C. In consonance with its "concern" about future
^° I technological developments rendering obsolete an entire product
if i line, the task force called for accelerating development of the
-S ; t model at the low end of the line, the model A, which was to be a
15 |j 360-compatible processor targeted between the 360/20 and 360/30, to
20 ;> take advantage of the "large and barely exploited market for a low-
i)
21 :j priced scientific computer". ( Id. , pp. 2-3.) The task force
]
22 ;} observed, however, that:
22 •} "The announcement of Model A will have an effect on
I the whole product line, all the way up to the 110 8A. Regard-
2^ j less of what is claimed, the fact that model A contains the
360 repertoire will tell the world that our other products
may be dead ends". (Id. , p. 5.)
-481-
i
*•!
i
3
7
3
Q
*
2
3
.4
5
,3
.7
d» The £000. Series . The task force had been convened
to consider Univac' s product strategy nearly three years after
IBM's SPREAD Committee report,- its reports appeared nearly a year
after the announcement of System/ 360. (DX 16, p. 1.) Univac
finally announced its third generation compatible computer family,
the 9000 series (corresponding to the previously mentioned models
A, B and C) in the spring of 1966. Called a "line of small and
medium-sized computer systems", Univac' s initial offering included
"the 9200, a low-cost, internally programmed punch-card system, and
the 9300, a high-performance card and tape system". (DX 70, p. 9.)
While the 9000 "aimed at compatibility" with IBM's 360 (Eckert, Tr.
908) , it was not truly compatible:
"[A] new line, compatible with IBM 360 coding . . . would have
probably solved the problem. While the 9200, 9300 and 9400
are IBM like in, their order code, they are not enough alike
to do us any real good. We have had loads of people prove
to us why we can't be IBM compatible and very little real
effort to be IBM compatible, either in our software or our
hardware efforts." (DX 10, p. 1; see also McDonald, Tr.
3803-04.)
The 9000 series was upward but not downward compatible
•* ,j among the three models. Thus, "if a person had programmed something
— ;j for some of these smaller machines he could use it in one of the
t
"6 ;j larger machines but not the other way around." (Eckert, Tr. 906-
i
" ] 07.) It also was not compatible with the 1100 series. (Eckert, Tr.
4
22 j 908.)
- I
'-* j The third machine of the line, the Model 9400, was first
i- ! announced in January 1968 (DX 13914, p. 6) , and delivered in 1970
Zz j "from factories in the United States, West Germany and Japan". (DX
!
■! -482-
i
I;
i
f I
i
!
3271, p. 7.) The 9000 series was intended to "enable smaller
companies to benefit from the advantages of computer power. . . .
Typical customers [were] a savings and loan association in Kansas
City, an aviation company in California and a wholesale grocer near
Philadelphia. " (DX 13914, p. 16.)
Univac both manufactured its own peripherals and purchased
7 ij peripherals from others, remarketing them as part of its computer
S
9
10
11
12
13
14-
systems.* For a short period, it marketed to other manufacturers
its peripheral devices which in turn were remarketed as part of
other systems.. (McDonald, Tr. 4053-55.) Further, its own products
were used as part of systems in another way. The 9000 series, for
example, were sometimes used as terminals to other manufacturers'
systems. (McDonald, Tr. 3969? Withington, Tr. 56981.) As a 1970
jl Univac advertisement said:
q. -J "They are widely used as either central site systems or
terminal systems. As terminals they may be upgraded, without
+ s >.i reprogramming , in low-cost steps to grow with your processing
lS 'l needs." (DX 13939 , p. 176.)
U jl In addition to acquiring peripherals from other manu-
-S ; f acturers , Univac contracted with software houses to have work done
*s : j when it did not have sufficient in-house capability to meet its
:|
20 j requirements and did not wish to expand internally to meet a peak
t
T5 H
* For example, Univac purchased tape drives from Ampex (Ashbridge,
23 | Tr. 34851) and disk drives from Memorex, Calcomp and Peripheral Systems
j (Guzy, Tr. 33170-71; PX 5584, p. 16; DX 1302, p. 1.) In 1968, IBM
24- i employees reported that Univac also purchased disk drives from Vermont
J Research, Bryant and Data Disc and tape drives from Potter Instruments
25 | and OKI. (PX 2267B, p. 27.)
i
I -483-
load. (Eckert, Tr. 915-16.) McDonald testified that Univac purchased
"software assistance from the Computer Sciences Corporation and also
from University Computer Company". (Tr. 4024.)
Univac- both leased and sold its EDP equipment. McDonald
rewrote in 1967 thatt
I :
jj
Ti
T.
Z
m
4,
5
6
7 i:
s
M [approximately 50 per cent of the Division's products are
sold outright with the remainder leased by customers on a one-
year to five-year basis." (PX 1, p. 3.)
Univac provided support services to its customers as well. (McDonald,
Tr. 2893-96.) McDonald testified that Univac had to provide these
services if it "were to compete successfully", since IBM did so.
(Tr. 2895-96.) However, Univac did not unbundle when IBM did in 1969,
because:
"[W]e felt that there would be considerable anxiety in the
marketplace . . . and we felt that it would be to our
competitive advantage to maintain our previous pricing
policy . . . and I think this was effective, at least for
a period of time." (McDonald, Tr. 2896.)
McDonald testified that Univac' s pricing policy between
1963 and 1971 was "to provide the potential customer with a system
that would perform his requirements at a price that would generally
be 10 percent, as a rule of thumb, below the price offered by IBM",
:{ not taking into account the performance of associated peripheral
.1 devices. (Tr. 2883-84; 4190-91.) Univac attempted to set its
~ :! •
\ products' price/performance between IBM's products, much as RCA had
'? i
^ : |done with its Spectra series. (McDonald, Tr. 4132-83.) Consider-
ing that Univac 's 9000 series was announced two years after Sys-
2- i
jtem/360, Univac' s pricing approach was perfectly understandable.
IBM was not the only competitor about which Univac was
25
i -484-
i
1
3 _;■ industry" [IBM, Univac, CDC, RCA, G£, Honeywell, Burroughs, and NCR]
concerned, however* While McDonald, in 1967, identified "eight
major hardware manufacturers" who were "[a]t the hard core of the
(PX 1,'pp*. 6, 12; see also McDonald, Tr. 28:04-06), he recognized that:
"[b]y the 19'60's, there were up to 50 major suppliers of
automatic computing digital and analog computers and data
processors. Over 700 organizations with some 30,000
persons were engaged in one part or another of the computer
field." (PX 1, p.. 1.)
These included peripheral manufacturers, software suppliers, service
centers, and leasing companies. ( Id. , p. 12.)
In the middle 1960s Univac management became "concerned"
about leasing companies. Forster wrote to McDonald in 1966, stating
that he had:
"some apprehension and also some prejudice in that I
consider them to be parasitic. ... If computers do
not stay on rental, since they have no loyalty to any
particular equipment their manner of disposal could be
damaging." (DX 78.)
Univac 1 s concern about the "manner of disposal" of leasing company
■ equipment was that the leasing company would at some later time
market it at very low prices, in effect "dumping it" on the market,
knocking Univac' s own equipment out of customer installations.
4-1
i
»
i
t\
8 [
5 i
10 ;
III
121
13
14- 1
15
IS
1? ;
I8 j;
id
j (McDonald, Tr. 4017; DX 76.)
20!
' Univac responded to this concern. In January 1969,
21 jj
] management approved revisions in Univac f s long-term lease plan which
:[ were designed to "decrease future vulnerability" to third-party
I leasing companies and which included the adoption of step-down
24 !
. payment plans for long-term leases and price-cutting of five-year
I lease rates for Univac ' s "most profitable systems". (McDonald,
•\
II -485-
i il
w
3
T
3
3
a
Z
A
£ !
LS
Tr. 3988-; DX 76, p. 5.)
e. Univac 's Success in the Late 1960s. Despite the fact
that Univac did not offer a single compatible family with the
breadth and compatibility of , the IBM 360, it experienced substan-
: tial growth in its EDP business during that time. At the end of
1965, prior to volume shipments of the 1108 or the announcement of
the 9000 Series, Univac 's U.S. EDP revenues were $203 million; as
of year end 1970, its U.S. EDP revenues were $478 million. (DX
8224, p. 624.) McDonald estimated in 1974 that "revenue growth
since fiscal 1965 [had] been 284%, or a 16% compound annual growth
rate — in excess of that of the computer industry as a whole".
(McDonald, Tr. 3867-68; DX 71, p. 7.) By fiscal 1969, the Univac
division had become "the largest contributor to [Sperry's] revenues
and earnings".* (DX 3271, p. 2.)
Univac' s growth was not limited to the United States.
From at least the 1960s onward Univac offered a single worldwide
product line. (Withington, Tr. 57602-03.) Thus, in 1967 Univac' s
International Division conducted operations through 32 subsidiaries
•J and distributors in Canada, Central and South America, Europe and
'} the Far East. Sales and service offices were situated in Belgium,
(England, France, Germany, Italy, the Netherlands, Norway, Sweden,
21 ] ■
j Switzerland, Australia, Brazil, Canada, Mexico, Argentina, Colombia
22 1
* Withington echoed the turn-around:
24
"The Univac Division became the largest single contributor to
?e 'I the profits of the corporation (it seems only a short time
~ I ago that Univac was castigated as the largest single drain
J on them!)." (PX 4834, p. 24.)
-486-
Li
Z
3
5
7
3
9
10
II
and Venezuela. (PX 1, p. 4; see also McDonald, Tr. 3839-42.)
McDonald predicted in 1967 that "overseas markets will grow at a
more rapid rate than that of domestic markets .... DNIVAC ser*
; iously intends to participate in the rapidly developing European
^ market'* and **wili routinely work across many international boundaries".
(PX 1,- pp. 5-6.) For 1970 Univac reported that its "international
t business [was] growing at an even higher rate than the domestic
operations". (DX 13915 r p* 7.)
Univac made great strides in the last half of the 1960s
despite its slow start in undertaking a compatible family of products
and its reluctance to accept disk technology. McDonald recognized
.-jj what the problem had been and what would be required to solve it:
£2 ,1 "Planning will be a requisite to survival on the basis
upon which profitable business development can be struc-
T ^ ; . tured. . . . The combined magnitude of both opportunity
i and risk superimposed upon the rapidly changing pace of
the industry will rule out success based upon 'seat of the
pants' decision-making. The old technique of fumble and
correct errors is out. There will not be time in the
future to recover from serious mistakes without suffering
;J severe penalties. We, therefore, must measure daily
17 ;| events against a flexible, preconceived plan of action
j in order to react in a timely fashion, competitively.
-S :[ Hard planning will be a part of daily activity. It will
not be a luxury in the future.
io
IS i|
:| "This is the precise area of one of UNIVAC s greatest
20 :j past weaknesses. It is an area which has received con-
centrated attention since 1964 and will continue to receive
emphasis in the future." (PX 1, p. 7.)
22 .1 Finally, Univac was back on its way to becoming a successful
22 j computer company .
2*1
! -487-
41. General Electric. At the time of IBM's announcement
of System/360/ General Electric was (as it still is) a large corpora-
tion. From, that time to the end of the 1960s, it was always in the
top six of the Fortune 500. ,(R. Jones, Tr. 8754.) Its corporate -wide
ii revenue grew from $5.1 billion in 1964 to $8.4 billion in 1969.
i{
t,| (DX 13667, p.. 1; DX 556, n. 2.) IBM's corporate revenue in 1964 was
r!| $3.2 billion (PX 5771, p. 3) and $7.2 billion in 1969. (DX 3364,
j i p. 5.) GE was larger than IBM throughout the entire period. (DX 556,
p. 28? DX 3364, pp. 59-60.) However, whereas most of IBM's domestic
revenue during the period 1964-1969 came from its EDP business (see
DX 3811; DX 3364, pp. 47, 48, 53, 54; PX 5771, pp. 32, 36; DX 13677,
pp. 33, 37; DX 13678, pp. 33, 37; DX 13679, pp. 33, 37; DX 13680,
pp. 45, 46, 53, 54), virtually none of GE ' s did. As the chart below
Ishows, at no time during the period 1964-1969 was GE ' s U. S. EDP revenue
9 \ more than 3-1/2% of its total U. S. revenue. (See Weil, Tr. 7260.)*
■0
!
si
7
S
*
1
L ;
1
I
5 ;j * Plaintiff called four witnesses who testified about the GE
] computer business. They were John W. Weil, who was in GE's computer
6 ;| business from 1963 through 1970, as Manager of Engineering
:i from 1964 through 1966 and Manager of Advanced Systems and
LI .!• Technology Operation thereafter (Weil, Tr. 7003/7007-08, 7072);
i John L. Ingersoll, who was involved with GE's computer business from
'2. :l 1967 to 1970 as a financial manager and a staff member of the Ventures
| Task Force (Ingersoll, Tr. 8042-43, 8097) ; Richard M. Bloch, who was
2 | Manager of the Advanced Systems Division of GE from November 196 8 to
; mid-1971 (R. Bloch, Tr. 7615-16, 7755, 7777); and Reginald H. Jones,
£4 j who held top management positions at GE beginning in 1961 and became
! Chairman of the Board in December 1972. (R. Jones, Tr. 8752-53.)
-488-
Li
z\
z \
1
4-i
5 !
6
7
3
9
10
ii i
12
13
14
15
16
17
IS
19
20
GE Total GE
U.S. U.S. EDP
Year Revenue Revenue
(in millions)
1964 $4011.5 $ 53.4
1965 4952.6 66.5
1966 5698.3 99.0
1967 6129.2 143.1
1968 6664.6 180.0
1969 6638.0 219.6
(PX 326 (DX 13668, pp. B, 3); PX 327, pp. 2, 36; PX 328, pp. 2, 23;
DX 556, pp. 2, 18; DX 8224, o~ 6.; DX 8631, pp. 31, 37; DX 13667, pp..
B, 14; DX 13669, pp. 3, 4; DX 14484, p. Rl; DX 7320.)
% GE
U.
S.
EDP
to
GE
Total U.S.
1.
3
1.
4
1.
.7
2.
.3
2.
.7.
3
.3
In 1963, computers were a part of the "industrial compo-
nents and materials area" at GE which accounted for 28% of GE's
revenues in 196 3. That area also included advanced controls for
machine tools, Lexan plastics, silicone chemicals, component motors,
appliance controls and lamp ballasts. The remainder of GE's business
j was derived from consumer goods (26% of revenue) , including appliances
! television, and lamps, among others; heavy capital goods (24%) ,
i including diesel electric locomotives and power generating and
21 il
22
23
24
25
transmitting equipment; and defense sales (22%) , including jet
engines and missile guidance systems. (PX 325, p. 10.)
Notwithstanding the small part played by computers in the
GE hierarchy, GE had to be considered one of the most significant of
I3M's competitors in the computer industry in the 1960s because, as
-489-
L ; Richard M. Bloch, who joined GE in 1968 as the Manager of the
;f
I ''■ Advanced Systems Division, testified, GE "was probably the greatest
l\ electrical and electronic technical organization, technically
M" oriented organization in the world, and with very strong financial
f | ■ resources " . (R. Bloch, Tr. 7615-17.) Similarly, in 1964 Withington
i :
5- ; wrote:. "GE's long— term potential must be considered greater than
j\ that of any IBM competitor". (PX 4829, p. 19.) John W. Weil, who
j i) was the manager of engineering for GE's Computer Department from
196 4 to 1966 and thereafter the Manager of GE's Advanced Systems
and Technology Operation until 1970 , testified that he believed
that "GE had the resources and technological capability to become a
major force in the computer industry". (Weil, Tr. 7007-08, 7072, 7173.)
With all that technological potential and financial power, GE was
called the "sleeping giant". (R. Bloch, Tr. 7788-89; PX 353, p. 43.)
But, in the computer field at least, the "sleeping giant"
never woke up. Its efforts in computers in the 1960s ended with the
sale of most of its computer business to Honeywell in the merger
that created Honeywell Information Systems. The story of how GE
failed to capitalize on its advantages and succeed is the story of
lack of corporate commitment, inadequate management and a failure to
keep up with the demands of the, market as technology and competition
advanced .
a. The GE 400 Series. During the year 196 3, GE was
l\ from NCR) and the 225. (Weil, Tr. 7005-06; see above at pp. 205-07.)
i
2-: In December 1963 GE announced its 400 series. (DX 488; DX 490.)
3 ;■ That series had evolved from work done in the Computer Department xn
4^1 Phoenix in the early 1960s. (Weil, Tr. 7238-39.) The 400 series
i
5 ! was called the "GE line of the future which would be compatible
!
S j throughout". (PX 353, p» 44.) The 400 line was aimed at, among
7 { others, IBM 1401 users. (Weil, Tr. 7031-35.)* According to Weil,
i
S i the GE 400 series (which was not compatible with the 200 series)
9- | "was intended primarily for business data processing users, although
10 : it did have some features that could support engineering and scien-
XI | tific calculations, but strictly as a secondary objective". (Weil,
12 i Tr. 7018, 7038.)
13 I However, within a few years after the announcement of
14.! IBM's System/360, "the distinction between a scientific computer and
I
12 \ a business computer . . . had been erased". (Weil, Tr. 7188-89.)
!
15 ■ GE was marketing the 400 for both scientific and business applica-
17 ! tions : "Can scientists and businessmen be happy with the same
^g || computer? Ask about a GE-400. Many installations have proved the
^g ; GE-400 can handle engineering and scientific problems as easily as
2q •! business problems." "So you see the GE-400 's don't just mean busi-
2t || ness. They now offer you the broadest capabilities available today
II
22 S
23 | * GE offered a "1401 simulator [with the 400 line] , a piece of
j software which . . . had some hardware assistance which permitted
24, \ programs from IBM 1401 [s] either to be run or to be converted easily
1 to the 400". (Weil, Tr. 703.1-32.
25 :
)
-491-
1
z
3
4
5
6
7
8
ia
12. i
13
on a medium scale information system — all the way from everyday
business runs to complex scientific problems. " (DX 489.) The
reason for this marketing change was, according to Weil, that " [a]s
of 1967, the [IBM] 360 had been on the market for three years and
the market in the middle range ... of computers was now much more
homogenized between business and scientific than it had been earlier,
and the GE 400 was hence sold as much as you could to a broad market
encompassing the middle class of . . . engineering and commercial
applications, both." (Weil, Tr. 7263.) " [S]o long as the scale of
problem is suitable to the machine the machine could do either
business or scientific work. The distinction between those two in
this class of machine had largely been erased by that time." (Weil,
Tr. 7264.)
74 <[ • According to internal IBM reports , GE also reacted to
IS ''
16
17
13
19 j
20 :
IBM's 360 announcement by reducing the price of the 400 CPUs between
8% and 17% and the tape drives and their controllers between 14% and
27%. (PX 2966, p. 3.)* The IBM Commercial Analysis Department
reported that "[t]he price reduction gives the GE 400 a price/
performance advantage over comparable System/360 configurations.
The improved price performance of the GE 400, coupled with 4-6
21 i|
;! * See also DX 1525, p. 1 (7/29/64): "GE has not officially
*— ;j reduced prices, but they are selling their 400 line at 18% off.
i They have also reduced their extra shift to a 10% charge"; and
23 ;j PX 320, p. 16 (6/23/64): "The 400 line is a competitive offering
.j today, but will require some revision if it is to remain competitive
*"* ■ in the direct access market, and in the mixed business and scientific
25
environment of two years from now. "
-492-
L
Z
I
•z
T
S
9
ia
11 !
12 j
13
14
IS
months delivery, demonstrable hardware, and programming support
makes the GE 400 extremely competitive with IBM's commercial product
line*" ( Id.- ) This Commercial Analysis report evoked disagreement
within IBM as to the effect of the GE price reduction. Khaplund
"felt that, while in some applications the price reduction did
indeed give the GE 100 [sic] a slight advantage, basically and
broadly the reason for the competitive announcement was that our 360
put them under pressure and they had to reduce the 400 for them to
stay competitive"* As a result, -IBM' s President, A.- L. "Williamsv
chided the President of the Data Processing Division, F. T* Cary,
for disseminating reports that were "unduly negative". (PX 2966,
pp. R-l, R-2.)
GE initially announced four models in its 400 line of
"compatibles"? in fact, however, only two were ever delivered. (PX
353, p. 44.) Subsequent GE product announcements (the 600 and 100
series) were not compatible with the 400 series. (Id.) In 1970,
is . ;}
'I GE cited the failure to deliver all the 400 models which had been
17 ;
is !
announced, as well as the incompatibility between 400 and 600 series
computers, as yet another reason why GE developed an "image of failCing]
13 ;
I to follow through" in EDP. ( Id. , pp. 43-44.)
20 '.\ —
f
I b. The GE 600 Series. GE announced its 600 series in
21. ] -
j July 1964, after the announcement of System/360. (Weil, Tr. 7197-
22 j
:| 98; DX 491, p. 1.) At that time the 600 Series consisted of the
j GE 625 and 635, which differed only in memory speed. Later, GE
[announced the 615, a "special configuration, slower memory speed
?c 1
-493-
L
Z
3
4*
5
6
7
a
9-
10
li
12
13
1*
IS
16 t
17 1
IS !
19 j
20 ;
version of the same 625/635 system", the 645, "associated with MIT
in Project MAC", discussed below, and, eventually, the 655 which
reimplemented. the 625/635 in higher speed integrated circuits.
(Weil, Tr. 7198.)
The GE 600 series also included "several compatible, but
physically different military versions". (PX 4829, p. 18.) In a
report on the 600 series, Withington wrote that the 600 series (and
the 400) show "the same design emphasis on well-balanced, practical,
but unspectacular systems. There are no technological innovations,
and their basic speeds and specifications are no more than comparable
to those of their competitors." ( Id. ) Internal IBM documents
reported that GE was offering the 635 "at no extra shift charge".
(DX 1525, p. 1.) When GE compared the 600 line against the announced
IBM 360, it concluded that "depending upon exactly which model and
details of usage and configuration, the 600 is either just a little
more favorable or just a little less favorable than comparable
members of the 360 series". (PX 320, p. 16.) As we have seen,
however, IBM had made its own analyses of the competitive reactions
to System/360 and improved its price/ performance with the 360/65
before delivery. (See pp. 389-90 above.)
2i j However, the 600 line was not as technically advanced as
22
23
24
25
the System/360. Weil classified the 600 series as a "second generation
solid-state computer". (Tr. 7192.) Moreover, in peripherals the
600 series suffered in comparison to the IBM 360:
-494-
L
Z
z
4.
f :
Si
7\
i
3- !
i
9 i
xa!
u
12
13
15
"At present, GE's systems are somewhat handicapped because
their peripheral equipment (particularly random-access file
storage devices) is in some respects inferior to IBM's. GE
says it is moving' actively to remedy this and to equal IBM's
peripheral equipment with products of its own manufacture . "
4829 r p." 1$.T
Nevertheless, according to Weil, the initial customer acceptance of
the 625 and 635 were "extremely good, well beyond our expectations"-
(Tr. 7206.)
One of the reasons for this was GE' s success with users of
the IBM 7090/70 9 4 computers* GE had "carefully targeted as one of
the markets for the GE 600 system the installed base of IBM's 7090 's
and 7094' s" because the 7090/7094 "was at that time by far the
leading scientific and engineering computer in the field, it had the
largest number of such systems, so it was a large enough target".
Further, since GE was itself a large user of the 7090/7094, the
"members of these computer installations played a leading role among
the user community of the 7090s and 7094s, so that ... we had an
ia I; enormous resource to draw on who understood that market and the
17 i needs of that user very well". (Weil, Tr. 7026-27.)
;|
•3 || GE "designed the 600 system to feel as familiar as possible
15 ;j to a 7090 or 7094 user". Among other things, its peripheral equipment
20 i could accept both media and format from such users and its software
21 ! represented "a compatible superset, a software that would include
•i
22 j| the capabilities of what the user already had but would give him
22 ii further extensions". (Weil, Tr. 7029.) To aid conversion, GE
24. "t provided a piece of hardware "called a 7090 Simulator, so that a
25 ! user who purchased this piece of hardware and put it in his system
-495-
I
2
3
5
S
7
9 !
i
t
u !
13
14
15
is ;
17
IS
could in fact run programs from the 7090 or 7094 without modification,
or at least that was the hope. Most of the time it succeeded."
(Weil, Tr. 7030.)
As a consequence,* when IBM announced its 360 line as
incompatible with its own earlier series, the computer group at GE
was "initially at least overjoyed with what had occurred because it
meant right at the time we were introducing a system designed to
3 I displace 7090*s and 7094's, IBM had itself abandoned the 7094 and 7090
computer series and brought out an entirely different computer
series, and it was our belief at that time that it would be easier,
if you were a user, to convert from the 7090/7094 to the 600 series
1£ I than it would be to convert to IBM's new 360 series. We regarded
that as a fortuitous occurrence and potentially to our advantage."
(Weil, Tr. 7060-61.) The user of the 7094 was "forced ... to
either go to a 360 or to some other competitive system, and we were
sitting there with a system designed to make that conversion as easy
as possible." (Weil, Tr. 7062.) That, of course, was one of the
risks that IBM was taking with the 360, and by 1964 GE with its 400
.Q '! and 600 and Honeywell with its 200 were attempting to take advantage
2Q : of the 360 's incompatibility with previous IBM lines.
2«f :j Weil testified that GE was "relatively successful in
22 ;| converting user programs from the 7094 and 7090 to our 600 line" and
2- |j that "our users found the conversion to involve work but to be
within reasonable difficulty". "I don't really think we found a lot
more difficulty . . . than we anticipated." (Tr. 7037-38.) Weil
24
S«5
-496-
L
a!
4*
5
i
a
ia
11 !
12
13
14
^ !
16 !
17 i
IS I-
estimated that GE acquired between 10% and 20% of the IBM 7090/94
base. (Tr. 7269.)
In addition to providing compatibility with the 7090/7094,
the GE 600 (as had the 400) provided a compatibility feature which
assisted conversion from the IBM 1401 to the 600 line. This would
enable users who had previously used 1401s as off-line devices in
conjunction with the 7090 or 7094 (e.g., tape to printer, peripheral
operations) to move both the applications previously done on the
7090/94 and the off-line functions run on the 1401 onto a single
computer in the 600 line. (Weil, Tr. 7031, 7034-35*)
The GE 600 series marketing strategy probably was based in
part on the ability of the 7090 users who leased the 7090 to terminate
their lease in a relatively short time and send the IBM equipment
which they were using back to IBM. (Weil, Tr. 7207.) In Weil's
judgment, "the GE 600 competed well with the IBM 7094". (Tr. 7212.)
Naturally, in competing for conversion of the 7090/94
customer as well as for other business, GE was competing against the '
newer System/360 IBM computers as well. Weil testified that "gen-
^g ij erally we were competing with the upper end of the 360 spectrum as -
2g ij it then existed. That would include occasionally the Model 50 but
2i ij primarily the various models of the 60 's and occasionally the 70 *s
22!; within the IBM 360 family". (Tr. 7207.) The restricted configuration
23 i| 615 may also have competed with the 360/40. (Weil, Tr. 7209; see
24
25
J also Tr. 7215.) Still latar, in about 1970, the 600 series competed
I
\ against the 370/145, 155 and (less frequently) the 165. (Weil, Tr.
-497-
I
z
z
5
S
7
S
9-
10
11 s
12 i
13
14-
15
16
7210-11, 7215.)
In "targeting" the 600 line against the 7090/94, GE in
part paid a price for its success . Weil testified that the GE 600
competed with the 360/65 "perhaps less well" than with the 7094
"because it was very specifically targeted at the 7090/7094".
(Tr. 7212.)
GE described its 600 line as a "family of large-scale
computers for business, scientific and real-time use" (DX 491, p. 1)
and as "a new, advanced, high-performance, large-scale computer for
use in business, scientific, and real-time applications — complete
with all software". (DX 492B, p. 3.) Weil testified, however, that
the line was originally intended "primarily for engineering and
scientific computation, but with specific features that would make
it attractive as well for business and commercial application,
but that in this case was the secondary market". (Weil, Tr. 7019.)
"While the machine was basically a scientific machine derived from
.- i the 7090/7094 we were trying to replace, we also included extensive
IS
19 !
character manipulation facilities , which would have been typical of
earlier business machines, commercially oriented machines, and was a
- n part of the support we built into the machine — there were other such
9T |j supports — part of the support we built into the machine for a COBOL
compiler to make it attractive to business applications". (Weil,
Tr. 7192.) GE "had the ability to use the growing low cost of logic
•i to provide a number of features aimed at these several markets"
;{ ( id. ) , reflecting the fact that "since the early sixties it really
22 |
23 i
24
25
-498-
L.
Z
Z
hasn't been economically important to design a computer system only
for business or only for scientific applications , except at the
extreme ends of this spectrum, where you were trying to do as much
scientific calculation as you possibly can within the limits of the
technology". (Weil, Tr. 7190.) Thus, GE' s "intentions" with
respect to the scientific marketing emphasis of the 600 series were
T |j differences of degree, not of kind. The perception at the time was
fl . j that the 600 series, like the 360, would compete in all application
areas.
Thus, Withington in 1964 wrote: "GE also believes (and we
agree) that in^the large-computer area there are no longer significant
distinctions between scientific and business machines, so the poten-
tial market for the 600 series and its successors is very large".
(PX 4829, pp. 18-19.) He also wrote:
S-
10
II !
12
13
14-
15
"GE's product line, then, is more analagous [sic] to IBM's
than that of any other competitor. GE hopes to compete not by
being different, but by doing the same things better: by
lo ;{ providing a combination of hardware, software , price , and
I customer service which will appear superior. No competitor
17 ; desiring a rapid increase in market share and profitability
'{ could afford to follow this approach. However, GE has repeatedly
IS ii stated that its intention is to build a solid and major position
i in the computer industry: its approach is consonant with this
IS ;! goal." (Idw p. 18.)
'I
2C J And Weil made clear that the 600 was marketed after its
21 ■] announcement for both business and scientific applications: "[A]s
22 "> the 600 was sold, as it went on in its lifetime, it was sold more
1
22 i and more to organizations that were more business-oriented and less
t
24, j scientific-oriented, partly as a result of bringing it down to the
:j
25 I 615 . . . which was more in the territory of more business installa-
1
i
1 -499-
L
2
2
4
5
6-
7
3
9-
ia
IX
12 I
13. j
IA
L5
17 1
19 il
20;!
tions, and partly because we found there were many more business
customers, out there than were scientific customers for our class of
system." "[Tjhe customer base that we built* up became more and more
business-oriented with time,." (Weil, Tr.. 7270-72; see PX 328,
p- 21.)
Weil testified that the 600 also had capabilities for real
time applications, which later turned out to be very useful for time
sharing (discussed below) although those capabilities "were used by
very few of the actual users that we sold the machine to". These
real time capabilities were "a direct reflection of the military
parentage of the central processor and the memory controller portion
of the 600 system". (Tr. 7192-93.) In particular, the development
of the 600 line drew on the work which had been done for the GE M-
236 military computer by the Heavy Military Electronics Department
in Syracuse.* (Weil, Tr. 7178-79, see Tr. 7301-02.) Among the real
tg !| time uses of the GE 600 were the data reduction and monitoring done
in connection with the Apollo launch system. (Weil, Tr. 7200; DX
556, p. 5.)
Even with the initial announcement of the 600 series, GE
was thinking about the importance of time sharing as an emerging
2-. :j area. In July of 1964, in an internal GE publication, the General
ii
22 !|
2^ ; * Weil testified that "for the hardware aspects of the central
; processing units", the componentry, skills and the manufacturing
i facilities required today to produce the central processing unit are
1 essentially the same whether one is speaking of a computer which is
used for scientific, commercial or process control application.
(Tr. 7191.)
-500-
25
L ■ Manager of the Computer Department was touting the 600's ability to
2 : permit "a large number of low-cost remote stations [to be] connected
3 ; . . . by common carrier lines, thus permitting many people access to
4- j the computer's problem solving skills. The need for many small
5 it computers on college campuses , large government installations , or in
a ■ widely-dispersed manufacturing organizations might thus be elimi-
7 || nated." (DX 491, p. 1.) While "the system had a number of features
3 1 in its peripherals and architecture which would make this possible
9
10
11
12
13
16 9
... at the time of this announcement we [GE] did not supply a
system that could support such an application". The hardware features
included "an excellent form of memory protection to isolate the
system software from whatever users may be doing and to provide
memory relocation features so we could accommodate a number of
different programs in the system at the same time". As we shall
see, the 600 was subsequently used in this way — as a time-sharing
system. (Weil, Tr. 7199-203.)*
Notwithstanding the attractiveness of the 600 series
17 s|
•I conceptually, GE encountered difficulties in delivering the 625 and
:t 635. Weil testified that:
t * As Weil explained, at the time it was thought that a single,
21 j central, shared computer was more efficient than a number of smaller,
i stand-alone computers. (Tr. 7203-04, 7254.) Thus, the 625 and 635
22 .; were "actively marketed for remote batch applications" as a cen-
; tralized system in which it was contemplated that remote batch
23 j terminals would replace earlier smaller stand-alone systems. (Weil,
| Tr. 7252-54.) In the 1970s with lower and lower hardware costs, the
2- ] trend turned the other way with many people believing that a number
■| of smaller computers were more efficient than a single large computer.
25] (See below atpp. 1276-86, 1339-40, 1448-59, 1510-16.)
-501-
"We were attempting to bring to market simultaneously a
new central hardware system, a new processor system, a new set
of peripherals, and an entire new set of software.
"On top of that this was the first time that General
Electric had ever attempted to put together and market so large
a system, and as a result of all of those factors at once, we
had a great deal of difficulty making the systems perform to
our customer's and our own satisfaction in the field. A combined
set of hardware difficulties and software difficulties",
including
" . . .a lot of difficulty with the magnetic tape units , we had
some unreliability in the memories we were using",
and, because of the size and complexity of the system,
\
} §i
"one of the difficulties we had was when something went wrong
J ; we had the problem of telling just what had gone wrong in this
roomful of equipment, so diagnosis was a problem for us as
well." (Tr. 7215-16.)
1 '. The difficulty with the software "centered around the operating
2 j system called GECOS, which was . . a comprehensive operating
4- supervisor", among the first of such systems. "[I]t was ambitious
in its design. We had a great deal of difficulty in getting it
5 • built, made reliable and made efficient." (Weil, Tr. 7216-17.)
7 ;J There were three versions of GECOS. GECOS I, which had
S I originally been intended for the 625 and 635, was never brought to
a j the field.
[Q j "It died in our test rooms because it was clear that it
| was sufficiently scrambled up internally that it would not make
rt .j a good product, and so GECOS II was constructed to take its
.1 place using the lessons that we had learned on GECOS I.
j "GECOS II was the first version of GECOS that was sent to
the field, and while it had a good deal of difficulty when it
went to the field, eventually, with much patching and baling
7i I wire, was made to operate satisfactorily.
LZ
15 \
-502-
1
2
3
5
6
7
a
9-
10
11
"GECOS III was initiated at that same — at the time period
that GECOS II was in the field again to make use of the lessons
we had in bringing GECOS II to the field, to reflect them back
in what we hoped then would be a clean design and a clean
product, so that GECOS III would incorporate the lessons of our
field experience.
"It was started and it was brought to the field much
later, I believe around 1968 . . . . " (Weil, Tr. 7 217-18.)
Weil echoed the theme of many computer people during the 1960s when
he said that GE's problems resulted at least in part because it was
attempting to develop a state-of-the-art software system. (Tr.
7217-19; see Perlis, Tr. 2001-04; McCollister, Tr. 9694-97, 9706-08;
Rooney, Tr. 12132-36, 12349-50, 12358; Conrad, Tr. 14088-89, 14133;
Withington, Tr. 56727-30.)
12 i The difficulties encountered with the 625 and the 635 did
13 I not result in slippages in delivery dates although Weil testified
14, ! ' that "perhaps they should have. The difficulties occurred much too
i
15 i often out in the customers' installations." (Tr. 7220-21.)
i
15 ' In late 1966 or early 1967, the 600 series systems were
tj ij withdrawn from the market and "put into . . . hibernation" ("it was
ia I put to sleep for the winter"). GE continued to support the systems
;i
^g : already sold but did not actively seek new sales. That winter sleep
2q '; period lasted for at least a year or two, and the systems were not
2i ij marketed again until 1968. (Weil, Tr. 7221-22.)
22 I; l n the fast-moving computer business, withdrawal is a
9«s '(mistake. Whereas IBM, when confronted with similar difficulties,
•!
24, I P ut a H °f its effort into solving them and keeping its customers
.1
3- 1 satisfied (see above at pp. 3 71-72 ) t GE withdrew. Weil said that
j
•i
:! -503-
.i
i
'i
1| "the hibernation of the 600 was a mistake"; "it led to a considerable
i
2 | undermining in the confidence of General Electric' s offering of this
i
3 ; class of system" and adversely affected GE's image in the computer
4-j industry. Weil explained why:
i
5 ]■ "When you buy a computer system, you are expecting a great
deal from the man who — the company that supplies it to you.
6 i You want to make sure that they will still be in business; that
they will stand behind any difficulties that your system has
7 j had, and that they will make it do what they told you it was
going to do. And any indications that people were backing away
8 | from such a full commitment would surely reduce a customer's
confidence in that particular vendor." (Tr. 7224-25.)* .
9
10 i
11
12!
13
15
15
17
13
19
20
21
22 'i ■ I
\ !
23 I
; * Withington, when he testified, emphasized that the customer's \
- A \ relationship with a computer systems vendor depended on the customer's |
"* ■' understanding that the vendor was "credible" — that is, that "a given j
manufacturer is a good one to be associated with . . . over time". i
(Tr. 55735-36, 57671-72.) i
i
i
-504- !
Withington also testified "that the inability of the
General Electric Corporation to deliver operating systems [including
GECOS and MULTICS] which completely met their advertised capabilities
hurt the credibility of General Electric." (Tr. 56754, 56727-28.)
There were direct financial consequences as well. In 1966
GE reported that "in the information systems business, current
operating losses were higher than projected because of difficulties
involved in meeting a very sharp increase in shipments, and because
of expenses in integrating worldwide product offerings. Substantially
increased costs were also encountered in getting some new systems
into operation." (PX 327, p. 9.) John L. Ingersoll, who had been
Financial Manager of GE's information systems business in the late
sixties, testified that from 196 5 to 1968 GE's difficulties with the
25
L
2
3
4-
5"
3 :»
7
a
9
10
II
12
LA '
15
IS
17
IS
19
V)
24. i
625 and 635 "were a major element in the financial results experienced
by that segment of GE". (Tr. 8339.)
Such difficulties , experienced with the first computers
and software of the 600 line, were aggravated when it came to the
development of time sharing.
c. Time Sharing. GE was involved in "two somewhat separate
threads" in the development of time sharing. (Weil, Tr. 7106.) The
first of these, developed by Dartmouth with some help from GE, was
"a very effective small time sharing system which we then brought
into our engineering organization and eventually modified, documented
and offered as a product . . . initially on a system derived from
the 225, later on a system derived from the 235, and eventually,
very related, conceptually related systems were offered on the 400
line and on the 600 line". Weil believed that this was the first
commercial time -sharing offering. (Tr. 7106-07.)
That system was "independent of the separate path which
it involved the more ambitious, technically, time sharing system based
;| upon the 645 and the MULTICS software". (Weil, Tr. 7106-07.) That
it
ij more ambitious development involved Bell Labs and M.I.T. (Weil, Tr.
■i
I 7108, 7225-26, 7231.)
,i
I Early in 1964, the Project MAC organization at M.I.T. ,
i
I which had already developed a time-sharing system (CTSS) on a pair of
I IBM 7090s (Brooks, Tr. 22739-40; Perlis , Tr. 1881, Weil, Tr. 7226-27),
1 was "interested in developing an extremely advanced time sharing
system". It approached a number of manufacturers "for a cooperative
-505-
L
m
3
x
7 i
3
.0
.1
2,
effort in that development". (Weil, Tr. 7108.) General Electric
proposed to Project MAC a version of the 635 system, which "would be
modified in accordance with some of the discussions we had had with
them, and which would provide then a hardware base for the advanced
time sharing system they wished to develop . " In addition , GE
"proposed working jointly with them in the development of the software
that would reside on that hardware." (Weil, Tr. 7111-12.)
In the summer of 1964, Project MAC selected GE over bids
from IBM, DEC (who did place a $1 million peripheral processor,
however), and others. (PX 2961, pp. 1, 3.) IBM believed that its
rejection was. due, at least in part, to the fact that it had proposed
to implement time sharing without dynamic relocation hardware.
(Knaplund, Tr. 90533-35.) Weil confirmed that GE believed that
"certain aspects of the 600 architesture [sic] , the 600 system, as
laid out, were more amenable to some of the things that MIT wanted
17 ;
** il
\
t
,^ i to do than were either the 7094 based system or the 360 based
system". "[W]e had a good meeting of minds, a good agreement on
It philosophy with the Project MAC team." (Tr. 7115.) Project MAC and
GE — and others in the industry — believed that computer systems were
evolving toward "an information utility" based on the time-sharing
20 '
i concept which would be of crucial importance to the future of com-
i puting. (Weil, Tr. 7116, 7251-52, 7254-55; see Perils, Tr. 2117-18;
22 "I
: PX 320, pp. 9-10.) Wright, Director of Time Sharing Marketing for
23
\ IBM from 1964 to 1965, summarized his eight or twelve conversations
24 I
II with Dr. Ivan Sutherland, the Director of Information Processing
?e 'i
|
■! -506-
Techniques for the Department of Defense's Advanced Research Projects
Agency (which funded Project MAC) (Tr. 13287-90) as follows:
"Ivan Sutherland was essentially exploring what IBM was
doing in the timesharing field. . . ' • I think that he was
trying to convince himself whether or not we were . serious,
whether or not we intended to follow through with a degree of
urgency in the project.
"He spoke words of encouragement, encouragement in the
fact that he believed that IBM should pursue development of the
time sharing concept in products and software as a matter of not
only great importance to the United States government, but also
of great importance to IBM and he simply encouraged and wanted
to be kept aware, sort of as an insider, of how things were
going on the project •
"[I]t was my understanding that his interests were the
fact that he believed providing time sharing facilities to the
Department of Defense contractors in design" of new weapon
systems, and use in other things, including health systems and
so on, would, in fact, foster the use of computers, but, more
i importantly from his standpoint, would assist in the solving of
problems that these people in their research and development
activity were confronted with and the use of computers would
facilitate the solution of those problems at a more rapid rate
and, therefore, accelerate the advancement of technology.
1 !
i
*!
z i
5 I
*
!
Si
I
I
7 1
i
3 |
9- !
I
ia !
ii
12
12
IA
15
16
17
tQ !!
19 | As shown above (at pp. 417-36), many people within IBM
^ u ;| also believed that the time-sharing computer utility concept might
<"• ;! well be the wave of the future and failure to respond to competitive
;i
*— ;j thrusts in this area — especially by a competitor with the power and
--. it
^ j potential of General Electric — might relegate IBM to a secondary
24 j position in the future. Thus, for example, in September 1964,
ij
! responding to the loss of Project MAC and of other important accounts
-507-
"[I]t was clear that he felt that two large companies,
such as GE and IBM, pursuing developments in time sharing, was
beneficial to the government, was beneficial to industry and,
therefore, that he thought that was a good situation." (Tr.
13290-92.)
25
in the time-sharing area, Nat Rochester, a member of IBM's Time
Sharing Task Force, concluded: "There is much more at stake than
these few prestige accounts. What is at stake is essentially all
computing business, scientific and commercial. . . ." (PX 1194A, pp
2-3.) Two days later, the Research Group of the Time Sharing Task
: ! Force wrote:
"There is a very strong probability that the 'computing utility 1
will be the way of all scientific computing in a few years, and
a good possibility that it will capture a substantial part of
the commercial market as well. IBM cannot afford to overlook a
development of this scope. We are currently in danger of losing
all contact with the leading developers of this concept. " (PX
2811, p. R-3; emphasis in original.;
Similar thoughts were expressed by other groups within IBM. The
Scientific Computing Department reporting on "remote scientific
computing" urged:
"Certain accounts have already been lost. A small set of
.4 '.[ key accounts are right now in the process of evaluations
leading to computer acquisition decisions. For every such
case, decisions disadvantageous to IBM appear to be in the
offering. In quantity, such losses do not appear to be large.
,3 . In quality, they will have a tremendous impact upon a very
large market . segment.
.7 ;
.3 ■
;j "If we do not respond on the time-sharing requirements in
^g ; the near future, the time-sharing market will be largely lost to
I GE who has responded to this requirement. A large part of the
balance of the remote scientific market will also be in
sS
•i
j jeopardy." (PX 2964A, pp. 4-6.)
■j Wright put it concisely:
;j 'And all during this period of time, in general, the
,- ;j industry was in a state of agitation because time sharing
" - i appeared that it might indeed be a new wave of the future from
^ '[ the standpoint of computing facilities for a company or an
institution.
-508-
c-
i !
i
z\
z\
4-i
i
5" |
t
I
5 ;
7 I
3
NX
11
12 !
13
14
15 !
IS .
17 !
"So there was clearly, you know, an understanding that if
IBM for some reason did not respond to this particular require-
ment of customers 1 need, demands of the customer, it was very
likely that those customers might very well buy such capability
from somebody else.
"[T]he significance would be that IBM would lose business
and that part of the installed base that IBM had at that point
in time would disappear." (Wright, Tr. 12843-45.)
In addition to being in the forefront of the new wave, GE
expected two additional benefits from its work with Project MAC.
First, "it was an opportunity for us to work with one of the organi-
zations that was widely regarded as an advanced thought leader in
the field, hence, we hoped to benefit technically from that work,
but also because it was based upon 600 line hardware, even though it
was largely incompatible with the 625/635, it would nonetheless
provide a reflection on the 635 and 625 hardware in the minds of our
prospective customers, so that the customers would feel that the
machines they were buying were related to and that he might someday
look forward to growing into the kind of applications that MIT and
, g || GE were developing on the 645". Second, "it lent an aura of advance-
ment to the rest of our commercial offerings." (Weil, Tr. 7122-23.)
GE and M.I.T. were not the only participants in Project
i
[ MAC. Bell Labs was also to be involved in the development of the
19 ;
20 l!
21 ;
;! MULTICS system, "a system, hardware and software together, for
22 ;|
j| carrying out a very advanced form of time sharing, a multiple access
.1 to extensive system facilities". (Weil, Tr. 7225-26, 7231.)
The first GE system installed at M.I.T. was a GE 635,
21
25
-509-
L
I
I
f
3
7
which was "used as a development facility, but the project was aimed
... at developing the MULTICS system, and a part of the MULTICS
system was a special expanded version of the 635, which was later
• termed the 645-" The 645 involved "major extensions to the central
processor, primarily having to do with the way in which memory was
addressed and accessed." "[T]here were hardware protection features"
and a "high capacity input/output controller". "A very advanced
2 \ form of dynamic relocation was included in the 645." (Weil, Tr.
9
.0
2.
3
.4-
'3
7227-28.)
In the fall of 1965, GE announced the 645 as a product at
the Pall Joint Computer Conference. (Weil, Tr. 7233.) In December
it announced that it was working toward the "broad commercial availa-
bility" of the 645 system. (PX 326 (DX 13668, p. 15).)* However, within
a year of the December 1965 announcement, the 645 was withdrawn "because
we began to realize that what we had on our hands was a research
project and not a product. . . . We were attempting to do something
: that had never been done before, and, in principle, we might end up
L7 ;, I
j discovering that it was not feasible. As it turned out, it was hard |
LS ;i I
i and slow, but it was feasible." Weil described the GE 645 as "being j
19 | i
j in the research project stage" until 1969 or 1970. (Weil, Tr. j
20 \ ' i
I 7234.) j
El ! '
'. I
■J !
22 ;» !
:j * GE had already bid a version of what came to be the 645 to GM ;
23 i Research along with time-sharing software, graphic console and devices, j
j Neither the hardware nor the software existed at the time , and the !
24. i consoles were to be modified versions of those already being marketed
! for military applications. (Hart, Tr. 80284-87.) ;
25 ' 1
I
i ■
.! -510- !
L j
i
i
i
1
3 !
i
*•!
i
5 \
I
6 I
i
3
ia
11 !
i
12 j
14 j
i
15 :
|
16 j
17 j
is !
At the time of the public announcement of the 645, the
software had not been developed and the 645 itself was not in
existence. General Electric never offered the 645 again as a product,
(Weil, Tr. 7234-35.)
While the 645 was intended "to provide a top-of- the- line
prestige luster to the 600 line and to our other products, and also
7 J to be a prototype for future sophisticated time sharing systems", as
it turned out, "because of its lateness and its difficulty, it
represented very little to General Electric except a drain on its
resources" (Weil, Tr. 7236) although "some of the features that were
pioneered in the 645 have since appeared elsewhere". (Weil, Tr.
7237.) Despite the potential which it had for future success, GE
never put its principal marketing thrust on the GE 645. (Weil, Tr.
7236.) The 645 was never delivered, and Project MAC received a
system designated a "636". (Wright, Tr. 13375-76.)
Although Weil and others believed in 1964 and 1965 that
the MULTICS system "could be technically feasibly designed", the
participants in the Project MAC effort "underestimated the difficulty
jg ij of successfully developing MULTICS". (Weil, Tr. 7232.) "[T]he
2q i{ system operated in the way that [it] was originally intended about
2t ,j three years behind its own schedule." Weil testified that this was
22« a consequence, first of the difficulties of cooperation among M.I.T.,
2- j Bell Labs and GE, and, secondly, because "the technical task that
« ij was being attempted was extremely sophisticated and many of the
25
subjects were at the state of the art as it was then known, and it
took a long time to iron out the details of implementing some of
■I -511-
i;
2 I
3
S
7
3
10
11
12
13
these important features". (Weil, Tr. 7233.) Such problems occurred
in other state-of-the-art software efforts, including those of IBM.
(Perils, Tr. 2001-04; McCollister, Tr. 9694-97, 9706-08; Rooney,
Tr. 12132-36, 12349-50, 12358; Conrad, Tr. 14088-89, 14133; Currie,
Tr. 15704-06; Withington, Tr. 56727-30.)
d. False Starts . On a number of other occasions during
the 1960s, GE began development of product lines which were cancelled
or greatly reduced.
Weil testified that in the early 1960s, a series known as
WXYZ was in development in Phoenix. "WXYZ was a series of four
systems of which the Z was to be the most powerful." By the time
Weil became familiar with it, "only the X and the Y were under
serious development". After "considerable evolution", the X eventu-
al ally became the GE 400. "The Y was to be a rather sophisticated,
15
IS !
larger system, but it was cancelled at the end of 1962 and its place
in the market spectrum was eventually covered by the beginning of
the 600 project."' Neither the W nor the Z was ever delivered.
(Weil, Tr. 7238-39.)
!j Following the announcement of the 600 series computers, GE
_ '; considered a series of new product lines . An important event which
20 :
•j triggered these lines was the acquisition of overseas affiliates,
'■! the Bull Company in France and the Olivetti Electronics Division
22 ;
(later known as GE Information Systems Italia) in Italy. (Weil, Tr.
7239; see also PX 326 (DX 13668, pp. 3, 15); PX 328, p. 18.)
According to Weil, GE at that time was interested in
25
-512-
I
2
3
4-i
5
producing "a world-wide product line which would cover the main
portions of the product spectrum", and as a result a series of
product lines were conceived. (Tr. 7239-40.)
The first such line, the GE 100 line, was conceived during
the tenure of Dr. Louis Rader. Rader joined GE in 1964 as Vice
President and General Manager of the Industrial Electronics Division
j \\ and took over the GE Information Systems Division which was formed in
3
S-
10
II
12
1965. (DX 13668, pp. 15, 33.) The GE 100 line consisted primarily of
three sets of processors which were to be manufactured in Italy,
France and the United States and which GE intended to market through-
out the world. (Weil, Tr. 7240.) In 1966, Rader was transferred
from the Information Systems Division to become General Manager of a
„ ! | new division, the Industrial Process Control Division. (PX 327,
i. pp. 9, 33.) Despite the fact that several study groups recommended
proceeding with the 100 line, Hershner Cross, who took over from
Rader as General Manager of the Information Systems Division in 1966
"overruled all the study groups and decided that the 100 line would
be abandoned". Cross did this "at the same time that he put the 600
IS
IS
17
13
j into hibernation." (Weil, Tr. 7223-24, 7240-41; PX 327, pp. 9, 33.)*
IS I
•t
20 1
j * Weil testified that the GE Italian computer operation pursued
21 j the 100 line after the decision was made not to proceed domestically.
■j The lower members of the line manufactured in Italy had their names
22 'j changed several times and were brought t6 market originally as the
.1115 and later as "successive members of a moderately effective, low
23 [priced business system". (Weil, Tr. 7241.) The GE Italian opera-
tion pursued the 100 line despite Cross's edict because "they had a
2i | strong general manager". (Weil, Tr. 7242.)
.1
-513-
1 ; After cancellation of the 100 line, GE began to consider
i
2 J different new product lines.
3. I "Upon cancellation of the 100 line, one of the measures
that was taken was to initiate a study centered in France, but
4 ! with worldwide participation, to spec out a more advanced line
than the 100 line that would serve the same general purpose.
5 :
"This project, known as Project Charley, met in Paris for
5 | a period of a number of months , but nothing broader came out of
that beyond a book of proposed specifications
7 '
"At that point there were some management and personnel
g | changes in General Electric and it was about at this juncture
that John Haanstra came to General Electric, and he initiated
g. i the development of another line of computers, again to be
worldwide and again to serve a broad spectrum.
10 i
11
*2 i The ERW line began in late 196 6 and "lingered on for a while after
that, but its principal effort was for eight or nine months, beginning
"Eventually this line of machines was known as the ERW
line . . . ." (Weil, Tr. 7242.)
13
14
15
16 .
17 !
13 ii
» 1
in the fall of '66, into the spring of '67". After that,
"John Haanstra 's responsibilities were changed and he was
put in charge of the Phoenix operation. He lost his personal
identification with this worldwide product line and instead
became a champion of what was going on in Phoenix, which of
course was very heavily the 600.
"The ERW line was largely leaderless for a period of time
:} "Then Dick Bloch came to General Electric and he instituted
l| a line, I believe initially called the 700 line and eventually
21
22
23
24
called APL, which was his conception of a worldwide, broad
spectrum computer line." (Weil, Tr. 724 3.)
This was in 1968. (R. Bloch, Tr. 7615-16, 7755.) Haanstra, who had
been recruited from IBM to lead the GE computer operations in 196 8,
was moved to Phoenix less than a year later, then was killed in a
plane crash in 1969, and none of these projects ever resulted in
-514-
I
2
3
4
5
6
7
3
9-
10
11
12
13
14
15
IS .
17 !
delivered products. (Weil, Tr. 7242-46; R. Bloch, Tr. 7756.)
Bloch, who came to General Electric from Honeywell via
Auerbach Corporation in 1968 and succeeded Haanstra as General
Manager of the Advanced Development and Resources Planning Division,
testified that when he arrived, there had been "several starts in
the direction of an advanced product line". (Tr. 7592, 7611, 7757;
PX 327, p. 28.) According to Bloch, while there were "some very,
very excellent developments afoot" , the operation was poorly organized
and "one would have wondered how this would ever be put together
into a line". (Tr. 7757-58.) One problem was that there were
various development activities under way under different auspices
throughout the company. (R. Bloch, Tr. 7759.)
"GE previously was typified to me as a company of great potential
in terms of spot accomplishments in various areas — software,
hardware, new attacks, in concept and in hardware too. But the
real question was, how was it all going to be put together?
That was one side of it. The other side of it was that nobody
thought about the total plan, the total objective, what this
business data processing world was all about." (R. Bloch, Tr.
7759.)
Thus, Bloch believed "the decentralized organizational
^ !» approach of General Electric adversely affected their attempt to
10 J
l * || develop an integrated line of computer products" , and individual
20 ; j departments took over responsibilities for obsolescent lines, promoting
21 ;j their own interests. (Tr. 7759-60.) The problem was that GE,
22
23
24
25
unlike IBM, was unable to tie together under central control this
disparate collection of products produced and marketed throughout
-515-
the world.* What Bloch called GE ' s "decentralized organizational
approach" was a substantial part of its downfall.
e • The, Management of GE ' s Computer Operation . General
Electric encountered substantial difficulties in managing its computer
; operation. It had a revolving door of management personnel running
its computer business during the period 1964-1970. During that time
frame GE ran through a progression of managers and other key per-
sonnel whose jobs constantly changed and who were succeeded by
people with little computer experience.** The result was that
^ i
f I
I
I \
|| * The following statement from the 1968 GE Annual Report summarized
the GE computer equipment business as of that date.
7 { "On a world basis, General Electric offers five product
lines, starting with the small-scale GE-50 series, produced by
Bull-General Electric in France. The Italian operation produces
the 100 series, and in 1968 introduced a more powerful system,
the GE-130. The US-designed 200 series continued to hold wide
acceptance for its dependable computing power. The GE-400 line
of medium-scale systems is produced in the U.S. and France and
in Japan under a licensing agreement." (PX328, p. 18.)
'.[ ** In July 1963,. the Computer Department was headed by Harrison
7 i Van Aken, who reported to the General Manager of the Industrial
I Electronics Division. (Weil, Tr. 7085; PX 320, pp. 1, 2, 19; DX 485;
3 ;jDX 491.) In 1964, Dr. Louis T. Rader was hired to be in charge of
ijGE's worldwide computer activities. He was named Vice President and
g ,, General Manager of the Industrial Electronics Divison and reported
:|to Hershner Cross, Vice President and Group Executive of the Indus-
g j trial Group. (J. Jones, Tr. 79357-58; DX 13667, p. 31.) In
41965, General Electric realigned its organization and formed an
£ l| Industrial and Information Group headed by Cross. Within that group
~ j w as the Information Systems Division, headed by Rader. (dx 13668,
"? :',PP» i ^' 33.) i n 1966, still another new division was formed, called
— jthe Industrial Process Control Division, and Rader was transferred
~ 'from the Information Systems Division to the new division to become
tits General Manager. Cross remained group executive of the Industrial
M ; |and Information Group and at the same time served as Acting General
"~ {Manager of the Information Systems Division. (Weil, Tr. 7223-24;
j- ;PX 327, pp. 9, 33.) Early in 1967, J. Stanford Smith, formerly Vice
-516-
2 !
3 !
projects begun were abandoned and no continuity of purpose or product
development existed. But the problems went deeper than that.
According to Weil, one of the "major mistakes" which GE
• made in managing its computer business, stemmed from GE's "very
strong" belief
7'i
3
9
10
II
"in the philosophy of professional management. This basically
is that management is a profession and a good manager can
manage any kind of business .
"This in fact works quite well for a mature or gradually
declining business, where a man put into a business can model
his behavior upon that of his predecessor's and then make
adjustments as he learns what's really going on. In a rapidly
evolving business, however, his predecessor's behavior, especi-
ally if it was unsuccessful, is a very poor model. And since
he knows nothing about the business , he is a professional
manager and came from Toasters or Welding, or whatever it may
. be, elsewhere in the General Electric Company, he really could
•^ If: not understand what he was managing.
13
-*" ij President of Marketing and Public Relations Services , became the
General Manager of the Information Systems Division. (DX 13668, p. 33;
PX 327, p. 33.) In January 1968, GE again changed its organization,
going from 5 groups to 10 groups and from 29 divisions to 46 divisions.
^ :! Hershner Cross's Industrial and Information Group was split into two
jj groups. Cross remained Vice President and Group Executive, heading
17 ';{ up the Industrial Group which included the Industrial Process Control
;| Division led by Rader. Smith was promoted to Vice President and
IS -[Group Executive in charge of the Information Systems Group, and John
j Haanstra, who had recently come from IBM, became General Manager of
13 -I the Advanced Development and Resources Planning Division within that
Igroup. (R. Bloch, Tr. 7755-56; DX 13669, pp. 27-29.) In 1968,
2£ | Haanstra became General Manager of the Information Systems Equipment
: l Division (PX 328, p. 28) and Richard M. Bloch replaced Haanstra as
21 [General Manager of the Advanced Development and Resources Planning
.| Division (later the Advanced Systems Division) . (R. Bloch, Tr.
22. :! 7623-25; PX 328, p. 28; DX 556, p. 30.) In early 1969, Hilliard W.
J Paige, who was Vice President and Group Executive of the Aerospace
23 [Group, replaced Smith as head of the Information Systems Group. (PX
1 328, p. 27; DX 556, p. 29.) Later that year, J. F. Burlingame
24. 1 succeeded Haanstra, who was killed in a plane crash, as Vice President
jof the Information Systems Equipment Division. (DX 556, p. 30.)
-517-
L ;
i
zi
i
3" !
i
I
s i
i
7;
i
3 I
9-1
10 |
11 '
12
13
15
16 x
17 !
i
13 ;
19 I
20 :
21 ;
22 ;
23 !
2± !
"But if you have a series of these managers above each
other, they feel they are in trouble, they now must do some-
thing. What can they do? They do not understand the business
well. So the only thing they can do is to replace the man
working for them.
"So the net result of this was, as we got into difficulties-,
especially in bringing the 600 to market thereafter, we had a
sequence of people running General Electric 's computer business,
none of whom, except when we come to Dick Bloch and John
Haanstra — and, again, they were not in charge of the computer
business but were key people — none of whom were experts in the
computer business. Furthermore, we had a new one every eighteen
months or so.
"So that General Electric never developed experienced
management that understood the computer business, and I believe
this was a major part of why General Electric never learned how
to manage the business properly." (Tr. 7247-49.)
This philosophy led. to General Electric having "a great
deal of difficulty ... in entering dramatically new fields",
1 a i although it was "extremely successful in managing mature businesses
and declining business". (Weil, Tr. 7259.) This is undoubtedly
! associated with GE's decision, discussed below, to remain with its
"core" businesses rather than continuing in computers.
GE's management problems were perceived outside of GE as
well. Withington testified:
"I recall that General Electric, consistent with its policy of
rotating managers between divisions , changed the senior manage-
ment of its computer systems business at intervals of approxi-
mately three years and I recall feeling that this was a poor
practice as the incoming managers rarely understood much about
the business at the time they would take it over." (Tr.
56731.)
John Jones, of Southern Railway, testified that although
Southern Railway was a "very large customer of General Electric" in
-518-
1 1
i
*i
i
4-1
5 I' ment that I was in at the Southern Railway Company and with the
s-i
i
t\
s !
i
11!
I
i
12
13
14
15
16
17
13
19
20
other areas, Southern Railway "did not seriously consider their
computer equipment". (Tr. 79352.) In the middle to late 1960s,
Jones reached the conclusion "that General Electric was not a viable
competitor, not one that I would consider selecting in the environ-
project that I had before me to complete". (Tr. 79353-55.) He
testified:
"[Tjhrough personal knowledge at several levels in the
company, I was at least to some extent aware of the activities
of the Computer Division of General Electric, and it was my
view that there were some serious problems in terms of how they
were managing that function, and it was my concern that I would
not be able to obtain the support and continuing responsiveness
from General Electric that I would judge to be critical in the
system that we were considering installing.
j "As a result of those concerns, despite the fact that we
had been a large customer of General Electric in other areas,
it was my conclusion that I did not want to take on the risk
of, or what I perceived to be a risk, of considering installing
General Electric equipment." (Tr. 79354-55.)
Jones' views crystallized in middle to late 1967 ( id. ) , about the
same time that General Electric' s difficulties were being made
public in the form of the "hibernation" of the 600 system.* (Weil,
Tr. 7221-22.)
j In 1970, GE ' s future product plans (then known as APL)
j recognized that among the "negative factors" which affected GE * s
ii
ji image in the computer industry were (a) GE * s management indecision
\ and replacement", (b) GE ' s "professional manager" image, (c) GE's
23
24 '-l
25
! * Jones based his views on his personal contact with General
Electric at the time. (J. Jones, Tr. 79355-60.)
-519-
l\ "lack of long-term commitment" and (d) GE's "loss of key personnel".
Z | (PX 353, p. 45.)
3; I Rotation of management meant a lack of continuity in decision
4| making over time, but another difficulty, as already observed, was
5 | decentralization of decision-making responsibility at the same point
in time. Bloch testified that
6
7[1
3
9-
ia
ii
12 1
13
14
15
IS s
I!
17 l
IS
i
19 j
20 ;
"GE operated in a decentralized fashion, with profit
centers usually at the departmental level, and for reasons which
I do not pretend to comprehend, the top management of the
company allowed these growths to occur of quite competent, in
their own right, groups, both here and overseas.
"Overseas, of course, one can understand some of that,
because there was outright acquisition. But even here there
were a multiplicity of centers and there was a proliferation of
activity; multiple peripheral devices of the same general
character being developed at different places at the same time;
a lack of coordination from any central area whatsoever.
"Our plan was, indeed, to make use of the facilities
worldwide but to have it completely controlled and specified,
all standards set, from the central operation in New York. And
this was a new philosophy to them entirely. And if this was
indeed a new philosophy to them, then I can understand why they
had problems earlier." (Tr. 7646-47.)
Bloch had "no question" in his mind that "the decentralized organi-
zational approach of General Electric adversely affected their
attempt to develop an integrated line of computer products". He
encountered "substantial resistance" to his attempt to limit the
:i decentralization. (R. Bloch, Tr. 7759-61.) GE's decentralization
21 || .
ji of responsibility within computers reflected its general management
j| philosophy. GE's 1968 Annual Report stated that "General Electric
23 >
s had 170 decentralized operating departments focusing on separate
:! aspects of world markets in 1968. Its production ranged over seme
25 ;
-520-
1
2
3" j. * The Computer Department was always buried deep in the
4~
3
6:1
3
9-
10
11
12 !
i
13. I
14
15
15 ,
17 I
3,000 different categories of products and 200,000 different models
and sizes." (PX 328, p. 7.)
organizational structure. Back in 1963, the Computer Department had
been within the Industrial Electronics Division which in turn, was
part of the Industrial Group. Weil testified that there were "some-
7 I thing approaching a hundred" departments in GE at that time. (Tr,
7153-54; DX 485.) In 1968, GE formed the Information Systems Group,
one of ten groups containing 50 to 60 divisions and, in turn, 130 or
140 separate departments. (R. Jones, Tr. 8794-95.) Because computers
had been so far down in its organizational structure and because it
had so many other products to attend to, GE failed to mobilize its
resources in computers to the extent necessary. Weil testified that
among the "major mistakes which GE made in the management of its
computer business" were two which related to this. First,
"a lot of ambitious and difficult tasks were attempted which
turned out to be more difficult and more ambitious perhaps than
was appreciated when we started.
Tg .! "Secondly, General Electric was never fully committed to
!{ its computer business. It was always a business . . . that
General Electric could live without. So that if troubles came
or budgets were suddenly bigger than had been expected, there
was always this reconsideration of 'Is this really a business
} we want to be in? And how do we prevent this from draining the
j profits of our other businesses? 1 It was not the strong commit-
i ment felt by those of us actually in the computer business of
^ ■■! General Electric." (Tr. 7247-48.)
22 ; *
mmm i| He testified that there were differences, for example, between GE ' s
2^ ! '
:|
:i commitment to the computer industry and its commitment to the atomic
!
power business to which it "manifested a greater commitment to
19 j
21
25 ;
-521-
success". Nuclear power was regarded as "an adjunct to that core of
business of the company" consisting of the supplying of power genera-
tion equipment. "It was clear that the mission of the nuclear power
business was: We don't know whether there is a business, but if
r there will be a nuclear power business, you will be one of the
leading competitors." On the other hand, the "equivalent charge"
for the computer business would have been, "We are sure there will
be a computer business, now you must demonstrate that you can compete."
(Weil, Tr. 7174-76.)
Similarly, Bloch testified that, when he joined GE in
1968:
"They were in the business. They had been in for some
period of time furnishing general purpose computing equipment.
My feeling was, however, that it was always tainted with some
tentativeness or speculativeness on the part of the company as
a long term commitment to the field. My feeling was that if it
turned out to be a great success, the company would be delighted;
if it turned out not to be a great success , the company could
extinguish parts or all of its activity in the field without
necessarily any great remorse." (Tr. 7623-24, 7616.)
! | Reginald H . Jones , Chairman of the Board and Chief Executive *
■'}
'I
! Officer of GE at the time of his testimony and a member of the
'J
Ventures Task Force which recommended GE's exit, as described below,
i [
j testified that he (and his predecessor, Fred J. Borch*) had agreed
i with their predecessor, GE's former Chairman, Ralph Cordiner, who
\ said about GE's computer business: "General Electric 's mistake was
3 ;
| * Fred J. Borch was Jones' predecessor (R. Jones, Tr. 8752), not
i I his "successor" as Jones mistakenly says at Tr. 8870.
-522-
L
2
3
a
7
3
1C
£1
12
12
12
15
17
13
19
that it failed to realize the opportunity and therefore made an
inadequate allocation of resources, both human and physical, to the
business". (Tr. 8869-70, 8731, 87520
Jones testified that "as early as the 1950 's, if we had
increased substantially the technical manpower assigned to the busi-
ness , if we had increased at that time the financial resources
required for the business , they would have been much smaller in
terms of absolute numbers than they would have been, let's say, some
fifteen years later." (Tr. 8875.) Jones said: "We never did make
the allocation of resources to the business that were warranted. "
(Tr. 8874.) The contrast with both IBM's commitment to the business
in the 1950s and its investment and risk-taking with System/360 in
the 1960s is striking.
f. GE's Position in the Late Sixties. In 1964,
General Electric obtained approximately half of Compagnie Bull
General Electric and Societe Industrielle Bull General Electric
for $43 million. • (DX 13667, p. 16.) By the time of the Honeywell
merger, revenue of the Bull companies was $206 million.
(DX 554, p. 10.) In 1965, General Electric acquired the majority
r interest in Olivetti -General Electric for about $12 million. This i
!
23.,
i subsidiary was formed from the electronic data processing business
21 ■!
\ and the electronics laboratory of Olivetti of Italy. (DX 13668, p. 3.)
22 ;;
i In 1968, General Electric changed the name of Olivetti-General
^ ;
! Electric to General Electric Information Systems Italia when it
I secured full ownership of the Italian-based computer affiliate . By
i
'I -523-
L
2
3
4
5
ei
i
7|
i
3 !
i
i
9-1
10 !
i
i
11 *
12
13
!
I*!
i
i
i
i6 i
17
IS
is ;
10
1968, General Electric had research, engineering and manufacturing
facilities at 13 locations in five countries with a worldwide sales
and service organization of 8,0-00 employees. (PX 328, p. 18.) As-
has been noted, the Bull subsidiary produced the GE-50 series and
the 400 series*" and the Italian subsidiary produced the 100 series
and the 130. (PX 328, p. 18.)' In 1968, General Electric also
broadened its line of input/output and storage devices and extended
its time-sharing services. By the end of that year, more than 50 GE
time-sharing systems were in place serving about 100,000 customers
in 17 countries around the world. GE reported that this area of the
business was "growing even faster than the computer equipment sector".
(PX 328, pp. 18, 21.)
GE also reported that "the company's investments in computer
technology have given us an expanded worldwide base in what has been
characterized as the world's fastest growing business. Again, our
developing capability to serve this industry is leading to further
j| new opportunities." ( Id. , pp. 3-4.)
ji
■ GE's 1968 sales of information systems were "well above
those of 1967 and with operating losses substantially reduced" .
( Id. , p. 18.) For the year ended December 31, 196 9, the General
I
i Electric computer operations which Honeywell acquired showed a
21 :i
;| profit. (Ingersoll, Tr. 8329-30; DX 554, p. 9.) Those operations
23
25
i| continued to show a profit for Honeywell in 1970. (DX 148, p. 1;
* The 400 was also produced in Japan and the United States. (PX
328, p. 18.)
-524-
1 1
3; J member of the large-scale GE-600 line" which had "had its best year
i
si
»
i
& I
t\
i
3 !
i
i
ii
12
DX 13977. )
In 1969, GE announced the GE-655, "the most powerful
in shipments and orders" . According to the GE 1969 Annual Report,
the GE-400 line also had a successful year. (DX 556, p. 13.)
Despite these improvements, GE was still in trouble. Yet,
if GE "did not appreciate the problem that was building in the late
Sixties" (R. Jones, Tr. 8876) , others did. Withington wrote in
1969:
"During 1968, General Electric was able to demonstrate completely
successful operation of its GECOS-III operating system for the
625 and 635 computers. . . . The 625 and 635 (recently joined
by a smaller 615) are continuing to sell largely because of the
success of GECOS-III, but the machines themselves are obsolescent
from the point of view of cost-effectiveness. It is to be
£3 j| presumed that General Electric has in development compatible
successor machines which can capitalize on GECOS-III, but which
2* | will show better performance. When this new line is announced,
General Electric will be in a position to make a strong resur-
gence in the large machine area." (PX 4834, p. 29.)
15
IT \
i
IS I
H
20 r
Withington judged the GE-400 line "obsolescent" as well and said,
"General Electric' s future position is dependent on the timing and
success of the new line." ( Id. ) As we have seen, GE had made
several false starts to the development of "compatible successor
machines" and was not, in fact, "in a position to make a strong
i
;j resurgence" .
|! As of 1969, GE had several incompatible lines, which had
i| been "developed at different times in different places, and to a
I
,| great extent under different management". (R. Bloch, Tr. 7787-88;
24 ;t
_ ; PX 3 28, p. 18.) Bloch, who came to GE in 196 8, concluded that the
25 1
-525-
L
2
3
4
f
6
7
3
ii
12
i*
15
15
17
IS
GE 100, 200 and 400 series computers
"were beyond their useful time in terms of state of the art.
They were in place doing their work, except that we were simply
facing the natural problem of the field, and that is with time.
You get to a point in which the price/performance is so improved
over equipment of days of yore that it is clear that those
users are going to move to new equipment, and either you are
going to provide that new equipment or your competitors are
going to provide it." (Tr. 7761-62.)
As Jones put it: "You had to bring out something that would exceed
the price/performance of the existing competition because you knew
full well that they were going to be moving ahead of you. It -is a
constant leap frogging game." (Tr. 8866-67.)
Bloch testified that although the larger 600 series had
come out more recently than the other lines, the importance of
smaller systems "far outweighed the significance of that 600 series,
looking toward the future". This was because Bloch foresaw a tendency
toward increasing decentralization and smaller processors, "which
are smaller physically, they are smaller dollarwise, but they cer-
tainly aren't smaller in terms of power when contrasted with the
earlier days". (R. Bloch, Tr. 7762-67.) Thus, while the 600 "would
!| be more appropriate for extremely large, powerful systems that are
19 i
! meant to be operated on a highly centralized basis", Bloch felt that
20 •.
'i GE needed to pay more attention to the smaller members of the line
| CR. Bloch, Tr. 7768; see Weil, Tr. 7252-54.)
22
24
25
The Ventures Task Force, organized in late 196 9 to consider
GE's future in the computer business, reported in April 1970 that
"most current product lines are obsolete" and that GE had a "lagging
technical position in mainframes, peripherals and manufacturing
-526-
Li
j
i
i
t
i
5 '
S-
7
8 . !
9-
10
11
12
13
14
15
16
™,
19 i
il
20 ;!
2Z !
process technology". (PX 331A, p. 18.)
This was not a secret held within GE. GE's reputation in
the computer industry had suffered badly from its management failures
and product obsolescence. Thus, the Advanced Product Line Master Plan
in January 1970 recognized that GE's image in the computer industry
was poor:
"General Electric has the reputation of the ' sleeping
giant 1 of the information systems industry, with vast capa-
bilities and resources which have yet to be marshalled for a
determined attack on IBM.
"GE's image is one of failure to follow through, as charac-
terized by:
" . An enviable image in the banking industry was
built through the success of the ERMA project and GE's
leadership in development of Magnetic Ink Character
Recognition standards. This image was subsequently lost
due to neglect.
" . In 1963, GE assumed a leadership position in the
area of communication systems and communications control
concepts with the announcement of the DATANET-30. Subse-
quently, GE has lost its leadership in the field by not
following up with any improvements until the recent announce-
ments in 1969.
**' || " . In the area of system capability, GE coined the
phrase, 'The Compatibles' . When the GE-400 line was
introduced, it was characterized as the GE line of the
future which would be compatible throughout. Although GE
announced four members (GE-425, GE-43 5, GE-455, GE-465) of
this line, it. delivered only two.
:| " . Since announcement of the GE-400 line, GE has
*• i{ made two other major line announcements: the GE-600 line,
which is not compatible with the GE-400 line; and the GE-
100 line, which is compatible , with neither. In fact, GE
currently supports seven mutually incompatible product
lines .
23
24- " . In 1964, GE recognized the way of the future by
an aggressive advertising and promotional campaign with
2- ; j regard to direct access. It indicated that direct
il -527-
F i
i
T\
3
?
J;
i
1 !
z
4.
access was the way of the future and announced a line of disc
storage devices to support this assertion. Since then GE has
not followed through on this commitment even though the initial
prognostications proved to be accurate.
"A brief summary of GE ' s image with respect to the various
product lines includes: *
"'GJyflMA 10 — an ideal model for a beginner.
"GE-50 — excellent for new users, but no compatible
upgrade •
"GAMMA 30 — an obsolete machine with no compatible
upgrade .
"GE-200 — an obsolete line with no compatible upgrade-
"GE-100 — a good family of products.
"GE-400 — a relatively obsolete line with no compatible
upgrade .
"GE-600— a reasonably good line with a need for a
higher member (a la the GE-655) . Good operating system
software — among the best in the industry.
"As long as the user is able to remain within a given one
of the seven product lines, he is reasonably satisfied.
"Measures of customer loyalty appear to fluctuate from
,6 jj year to year, but are generally below IBM and appear to be
•j consistently below the industry average. This loyalty is
j ; understandably low when customers must move up from the product
i line which they are currently utilizing." (PX 353, pp. 43-44,
a ;l footnote omitted.)
( g ; ! Unlike IBM which had integrated its product line in 1964
?g jwith a single compatible line conceived, developed and marketed on a
•i
n .jworldwide basis, GE in 1969 still had several incompatible lines
w jwith fragmented development and inadequate worldwide coordination.
I
2 « 9' The Advanced Product Line (APL) . Plainly, if GE was
f
74 jto overcome its problems, it needed a new product line. This was to
?B ;be 31och's task when he joined the company in 1968. (R. Bloch, Tr.
i
;| -528-
L
2
3
5
S
71
8 j
i
9- I
I
10 i
i
i
11 !
i
12. j
13 I
14
15
IS t
17 j
IS j
i
20 ; !
7757.)
This new line, "initially called the 700 line and eventu-
ally called APL . . . was his conception of a worldwide, broad
spectrum computer line". (Weil, Tr. 7243.) It was to be "a single
integrated line to be marketed on a worldwide basis". (R. Bloch,
Tr. 7798; Ingersoll, Tr. 8104.) GE's plan was to achieve the "number
two position" in the field:
"We could not also see a company such as GE being satisfied
with a. $50,000,000 business, say, in some convenient corner of
the field, even if it were able to make a. profit there, which
might indeed happen, because a business that size is insignifi-
cant in the GE scheme of things." (R. Bloch, Tr. 7648-49,
7799.)
As a result of this goal, GE's APL was to be "an attack" "across the
board". The new line was to "attack" everything from the $500-a-
month rental to the $70,000- or $80,000-a-month rental, which, as
Bloch put it, "is a tremendous range", "well over 90 percent of the
total range". Of course, GE was attacking IBM and "in particular
attacking the IBM 360 series, and not only the 360 series, but what
we surmised was coming soon, and which became the 370 series." (R.
Bloch, Tr. 7647-48.) Bloch testified that, had the APL line ever
been completed, it would have been a "more ambitious ... or broader,
more comprehensive, line than any that was in existence in the year
21 ;! 1970 — or '69 . . . with the exception of IBM". (Tr. 7803.)
*— ; Bloch felt a sense of urgency. about this mission and sent
" j a telegram to various parts of the GE computer organization (including j
*■ ■ ! riaanstra in Phoenix and Weil m Bridgeport) creating a special task j
1 '
! force. (DX 540.) Bloch testified that he fait the situation was ■
' i
\ ' !
-529-
1 i
I !
I !
I ■ !
■i . (
25
1 i
i
2i
i
i
!
Z ;
i
4-1
si
i
I
Si
1
i
7
3
9-
10
11 !
12 j
13
14
15 !
16
17
IS '1 - .
j of anticipated worldwide shipments would be made to users of earlier
19 '•!
!| GE lines, with another 35% to be shipped to users of competitive
;i systems, chiefly IBM. The remainder were to go to new users. (PX
* I 353, pp. 53-54, 57.) To effect the necessary conversions, GE planned
\ to offer various emulation and conversion aids. (R. Bloch, Tr.
~~ !l
** 'j 7881-84; PX 353, pp. 53, 62-63, 67, 118, 119, 164-65, 171, 175, 178,
24 j 179.)
; i In order to induce IBM users to convert, the APL line had
;i
i
!j
! -530-
urgent because
"the company was on a timetable if it was to enter the field in
a fashion which I thought was necessary, which meant that we
had to fix the specifications, characteristics, and get the
assignments made, development of assignments and so on, through-
out the world, certainly before the end of that calendar year.
« •- •
"The importance was simply that of time costing the company
its future position in the field. By delaying the time at
which we could announce and ship these systems, we would be, it
was my feeling and generally agreed, losing some of our current
base.
" Secondly r IBM was, I thought, much more vulnerable at the
earlier time within this period, that is, in the earlier seventies
and that every month that could be compressed with respect to
the schedule meant an ability to tackle IBM more readily and to
preserve our customers . . . the present GE CPL [current product
line] customers who had obsoleting equipment. And there was
the danger, thus, of their moving elsewhere." (Tr. 7792-93.)
The APL line was not to be compatible with the earlier GE
lines. (R. Bloch, Tr. 7873; see also PX 353, p. 119.) Users of
earlier GE lines converting to APL would encounter conversion problems
of the same kind that an IBM user would in converting to APL — and
GE, of course, would have faced the same problem IBM was confronted
with in 1964 when it announced 360. It was planned that about 35%
s
7
3
la i
u
13
15
L| as primary targets, "the 360/20, 25 and, to some extent, the 30 and,
Z if to still a lesser, extent, the higher level machines in the IBM line,
Z\ and also another IBM line". (R. Bloch, Tr. 7663-64, 7866.) GE
4»| targeted ' these- users because of the difficulties that the users of
5" | the lower machines in the IBM line would have in converting from DOS
to OS/360 (R. Bloch, Tr. 7867-68) and the fact that most of the
programs written for such systems were written in higher level
languages. (R. Bloch, Tr. 7868-72, 7880; PX 353, p. 64.)
Bloch testified that he and "the top programming experts"
at GE believed that the conversion objectives of the APL could be
achieved, although it was "an extremely ambitious task". (Tr.
7889-90.)
To induce IBM's users to move to the new GE systems, a
price/performance advantage of 20 to 40 percent against the 360 was
thought to be required, and Bloch thought that it would be necessary
£g ;| for GE to match IBM's peripherals as well. (R. Bloch, Tr. 7654-59.)
yj I Basically, what GE was intending to do was to duplicate IBM's 360
t« :l plan of attack some five or more years after its announcement.
;i
t- 'J The strategy for APL preferred by the GE Information
•?
„- 1 Systems Group was the "A-F strategy" providing for the offering of
_? t
;! shipments beginning in early 1973. This .would have required an $858
„.. ! million expenditure before taxes, with an after-tax investment of
m \ \ $429 million for the years 1970-1975. (R. Bloch, Tr. 7695-96; PX
\ 362, p. 4.) Roughly half of the required investment was the financing
£•
?«
„, : the entire line at once, a "full across-the-board strategy", with
-1 i • !
-531-
1»
z
I
*
a
7
8
S
.0
.1 !
i
i
14- !|
re
LS ,
L7 [
LS
that would be required for the 80 percent of the APL line that was
expected to be leased by customers. As Bloch testified, it presumed
a successful. APL with a large number of systems on lease: "It is
one of the prices you pay for - success." (Tr. 7699.) "And from my
view, once those machines were out in the marketplace, we were going
to keep them out there, which meant tremendous income coming at a
later time." (Tr. 7929.) " [I]n no sense did I consider 858 to be
the exposure of the company." (Tr. 7703.)
Indeed, Ingersoll testified that, during the period when
he was associated with the Ventures Task Force, it was "a. general
assumption" that the announcement of APL would have the overall
effect of increasing revenue and income from GE's current product
line during the years immediately following 1969, with an increase
of $177 million from the combined product lines in 1970-1975 (Inger-
soll, Tr. 8378-82; PX 362, p. 10) , a positive effect not taken into
account in the $858 million investment estimate. (R. Bloch, Tr.
7935-36; see also' PX 362, p. 10.) Substantial net profits were
expected to be earned in the late 1970s, after which a successor to
•j
;j the APL was contemplated. (R. Bloch, Tr. 7908-10; PX 362, p. 5.)
1 The APL plan, then, was an ambitious one, requiring large
i expenditures. It contemplated an across-the-board attack, even
21 j
']• though profits might have been made in a "$50 million business"
J without such an attack. Further, it had to be pursued immediately.
y ■» i«
; In the event, it became just another false start.
24 f
I
t
1
•t
'I -532-
L
2.
<*■
4-
3
7
3.
9
ia
Li
121
13.
14.
15
16
I;
17 l
19 I
2a !
•i
'i
21 ;1
h. The Ventures Task Force and the Decision to Disengage.
The Ventures Task Force was formed by GE Chairman Fred J. Borch in
the last quarter of 1969. It was asked to review GE's computer
business , commercial jet engine business, and nuclear energy business.
Its mission was "to analyze those businesses and to present to the
corporate executive office some plans that would outline the alter-
nates and options available to the corporation with respect to those
specific businesses" . (R.. Jones, Tr. 8756-57.)
In particular, the Task Force stated in its report that
Mr- Borch "specifically impressed upon us the urgency of our finding
some way to arrest the heavy continuing drain on our assets resulting
from these major new ventures". (PX 331-A, p. 5.)* The Task Force
"adopted two broad criteria as the bases for our efforts to evaluate
each available strategy; the risks and potential rewards inherent in
each strategy and impact of each strategy on corporate earnings".
( Id. , p. 6.)
Corporate earnings were a problem. GE's earnings per
share had "plateaued" from 1965 to 1969, creating "a dismal record".
(PX 331-A, p. 5; see also DX 550, DX 551.) The 1969 earnings had
declined due to an "extensive strike". (Ingersoll, Tr. 8266-67; see
also PX 331-A, p. 5; DX 556, p. 3.) As a result, the GE stock price
j had declined 34% from 1965 through 1969 < compared to a decline of 17%
for the Dow Jones Industrial Average, 8% for Westinghouse , and an
1 increase of 3% for the Standard & Poor's Industrials. As the Task
24 1
;j Force put it: "Stockholder impatience is indeed understandable."
.'J * "Major new ventures" were distinguished from GE's "core"
| businesses. ( Id. )
:l -533-
L
2
5
a
7
3
9
LO
11
•2
12
(PX 331-A, pp.. 5-6.)
The members of the Ventures Task Force, Jones, Jack
McKitterick, and Robert Estes were corporate officers, but not one
of them had had any responsibility for the computer business or the
GE Information Systems Group.* (Ingersoll, Tr. 8267-69.) The
Ventures Task Force studied the companies that GE met in the market-
place only "in a superficial way" in order to understand the strategy,
types of equipment, and the "markets" attempted to be served by "each
of the major entrants in the business".** (R. Jones, Tr.. 8778-79.)
Jones testified that the Task Force was a "part time"
assignment for its members . He characterized the thoroughness or
completeness of the work done by the Ventures Task Force in the
following way:
r . j; "When you look back and think that we worked together for
" a very limited number of weeks, and when you recognize that the
computer industry is a very complex business, and when you
recognize also the fact that none of us had any experience in
, „ 'j the computer business before we went into this, certainly it
* B • was not an exhaustive analysis of the computer business. It
t-> ; was an analysis that I think developed a fair comprehension of
^ y \ General Electric 's position in the computer business, but I
wouldn't characterize it as an in-depth study." (Tr. 8767-6 8.)
Of the three new "ventures" — computers, jet engines, and
\ nuclear energy — the Ventures Task Force studied computers first.
1Q J
21 i
I * At the time of their Task Force assignment, Jones was Vice
22 ] President of Finance, McKitterick was Vice President of Corporate
:| Planning and Estes was Vice President, Secretary and General Counsel
23 \ (Ingersoll, Tr. 8267-69.)
2- t ** The Task Force studied IBM, NCR, CDC, Honeywell, Burroughs,
•I Univac, Xerox, ICL, Siemens, and several JaDanese companies. (R.
25 I Jones, Tr.. 8778.)
-534-
I
2 :
3
a
6
i
3!
i
10 |
11 !
12 I
13 !
14 I
i
15 '
16 i!
17 \
IS ;
19 :
This was because, as Jones testified, GE was "in a position in
nuclear where we had so many contractual commitments that our options
and our alternates were rather restricted. The same thing held to a
lesser degree in the commercial aircraft engine business, whereas in
the computer business it seemed to us that we had a good deal more
freedom to select from a rather wide range of alternates and options
as to the future course of the business". This was so because the
computer business "did not have long-range, long-standing contractual
commitments to deliver product [s] over an extended period of years".
(Tr. 8758-59.)
The Ventures Task Force "attempted to evaluate the risks
associated with the APL plans . . . from a broad business standpoint.
. . . [It] did not undertake to verify the accuracy of specific
details of the cost estimates, for example." (Ingersoll, Tr. 8431.)
The Task Force ultimately reached the conclusion that GE
should "disengage by combining its computer business with that of
some other computer manufacturer" (R. Jones, Tr. 8801) despite the
fact that it found that in the computer market "great size and very
rapid growth make for a challenging opportunity" with the U.S. and
„ ,1 European businesses projected to double in the next five years. (PX
£U i
:l
;i 331-A, p. 9.) It listed a number of negative factors affecting GE:.
•I "substantial operating losses", "heavy debt obligations and interest
22
23
24
burden", "obsolete product lines", and "poor reputation and image"
j (PX 331-A, p. 49), and stated that GE had
;| "Limited technical strength other than in data management and
25 ;j multi-processing software and communication equipment.
'!
";! -535-
I I "Major product lines obsolete, complete but incompatible.
Not vertically integrated. Weak in peripherals, mass storage
Z\ and terminals." (PX 371-A, p. 39.)
3 | Among listed "Critical Future Problems" were
i .
4 | "Across the board system obsolescence.
5 J' "Vulnerability of PARC [installed base] to competition — lack of
specialization. Customer loyalty now under 80%, lowest of any
$ ; competitor.
i
7J "Lagging technical position in main frames, peripherals, and
manufacturing process technology." ( Id. , p. 40.)
3 '
9
10.
11
12!
13
14
15
16 :!
■\
17 '
u •■'. that plan
* t " conceptually recognizes the current needs of the business and
Tq 1 presents a goal that, if realized, would indeed place the
:| company in a strong position in the business computer field.
:j It is our conclusion, however, that the APL entails very high
*° :| risks, and that it is doubtful that it could be kept to time,
21 ;j
ij
22 :j * Jones also testified: " [I]t is my experience that in business
«- :! you succeed when you satisfy a customer and when you do it in terms
2-* ;j of giving values that are highly satisfactory from the standpoint of
the customer. And I use 'value' in the sense of conveying reasonable
price, quality of product, features of product and performance,
overall performance of product." (Tr. 8 868.)
Its installed base was termed "already obsolete and vulnerable with
the conclusion: "Time is not on our side." (PX 331-A, p. 20.) The
Task Force stated that "we need to be realistic about the relatively
poor reputation and image we enjoy as a computer equipment manufac-
turer". (Idw p. 34.)
Jones concluded: " [W] e were not doing the job that was
satisfying the customer to the extent that certain competitors
were." (Tr. 8886-87.)*
The Task Force evaluated the APL plan. It concluded that
24
25
-536-
I
a!
I
i
4 |
si
i
I
5 \
7|
8'j
i
*!
J
lfl i
11 j
12 j|
i
i
13 I
14
15 i
16
cost and system performance schedules. Even if General Electric
were in a position to undertake such an ambitious program, we
would not recommend that it invest the requested sums in such a
hazardous project predicated on an all-out attack on IBM, one
of the world's strongest corporations.
"Faced with the lack of earnings growth, but seeking to
retain its image as a growth company, General Electric cannot,
in our opinion, undertake any half-billion dollar venture, such
as APL that produces substantial immediate net income losses.""
(PX 331-A, p. 7, emphasis in original.)
The APL plan, according to the Task Force, called for a fourfold
expansion in GE total shipments in six years with an expansion of
60% to 70% per year of its sales force and "even so, productivity of
GEC's] sales force must be twice as great per man as that of IBM".
( Id. , p-. 28.)
Jones testified that one of the reasons that the Ventures
Task Force "felt that the APL plan was one fraught with risk" was
that it called for technology beyond the current state of the art,
which required "invention by schedule in order to achieve its objec-
tives". (Tr. 8769.) If it "had not been so ambitious in a techno-
i
17 j| logical sense and in a timing sense it might have been a somewhat
l' it
IS
19
20
21
I better plan. It might have had a chance of acceptance." (R. Jones,
i
I Tr. 8790.)
i
| In fact, Weil testified that by late 1969, APL was not
i
I
| "well along" in its design and development and that "the software
!{ was still in fairly early specification form". (Tr. 7244-45.) At
ZZ : :
23
24
25
the time of the Ventures Task Force "detailed engineering specifica-
tions" and "firm cost estimates" were not available. (Ingersoll,
Tr. 8370; see also R. Jones, Tr. 8768; PX 363, p. 15.)
Yet the world would not wait for GE to come up with its
-537-
1
1
3
5
5
7
8
10
11
12
13
14
15
16
17
IS
19
20
360. Jones testified that "it was our experience that every time we
went out to sell a computer there were a lot of other people knocking
on the customer's door, attempting to sell him a computer. In that
sense it was highly competitive." (Tr. 8861.) He testified that
"it was the opinion of the Ventures Task Force that some of the
companies in the field would find it necessary to combine . . . that
the [business computer systems] business was competitive . . . and
that it would continue to be competitive and it would be those
competitive pressures that would cause some combinations to take
place in the field." (Tr. 8864-65.) General Electric believed
"that one of the characteristics of the business computer systems
business was that competition constantly forced suppliers to come
out with better products at lower prices in order to keep the custo-
mers that they had and to get new customers". (R. Jones, Tr. 8865.)
The Ventures Task Force called IBM "a moving target" which
Ingersoll interpreted to mean that "we at General Electric should
assume that IBM would not be a stationary object, it would be a
dynamic situation, and the conditions . . . might well change". "[I]t
was in effect a high risk to assume that the conditions and evalua-
tions . . . would remain constant, that is, that the comparisons
•i
•{ would be subject to change as IBM made plans and introduced its own
21 (j
|l products." (Ingersoll, Tr. 8128-29; PX 363, p. 15.) Similarly, the
22 :j
i Ventures Task Force and its support staff felt that "it was a high
23 |
■
I risk assumption" to assume that competitors other than IBM "would
24
; stand still with respect to market share". (Ingersoll, Tr. 8127.)
: \
'i
i -538-
L
Z
3
5
7
i
3 ;
-I
ii i
13 I
!
i
15 !
is ;
As a result, Jones testified that the Ventures Task Force
concluded "that the life of a family of computers was quite limited",
something in the range of four to six years, "and that you did not
bring out a family of products that simply met the price/performance
characteristics of the then existing competition. You had to bring
out something that would exceed the price/performance of the exist-
ing competition because you knew full well that they were going to
be moving ahead of you. It is a constant leap frogging game." (Tr.
8866-67.) It was a lesson which GE learned too late and IBM had
learned well before. (See PX 1077;* DX 4806;** pp. 493-94, 531 above
Indeed, the APL plan itself stated in January 1970:
"One of the key aspects of technology in the computer field is
its high rate of obsolescence. Never in the history of tech-
nology has the pressure of competition and the lure of highly
rewarding markets created such a dynamic evolution. While this
characteristic is forcing rapid technological progress, it is,
at the same time, imposing on the computer manufacturer a heavy
financial burden and the necessity of planning products with a
narrow margin for error." (PX 353, p. 23.)
The Ventures Task Force, of course, had not been called
•' jj together simply to consider computers. Ingersoll testified that he
IS
19
20
!(
* Thomas J. Watson, Jr.: "I believe that whenever we make a new
j machine announcement, we should set up a future date at which point
| we can reasonably assume that a competitor's article of greater
'•« capability will be announced. We should then target our own develop-
2 * 1 ment program to produce a better machine on or before that date."
:j (PX 1077, p. 2.)
-« ;, ** Thomas J. Watson, Jr.: "I think it important to note, however,
2,3 !J since we seem to have suffered for a few months or even years because
;i our machines predated the effective competitive machines now in the
2** ;| marketplace, that we now make these machines good enough so they
j will not be just equal to competition, for I am sure that once they
j are announced our competitors will immediately try to better them.
j This is all to the good and I am for competition, but I want our new
j line to last long enough so we do not go in the red." (DX 480 6.)
-539-
25
I !
believed that GE's management was concerned with improving the
profitability of the company as a whole, based in part on the expe-
rience of the late 1960s in terms of stationary and then declining
earnings. It was therefore concerned with the immediate impact of
APL on GE's earnings which impact, from 1970 to 1975, the APL plan had
projected to be negative. (Ingersoll, Tr. 8271-72.)
r !( The Ventures Task Force concluded that "General Electric
3 \ can ill afford the financial resources needed for an all-out drive
a for position in this industry, basically because of the needs of
. i other businesses within its scope". (PX 371-A, p. 73.) The "core"
businesses needed "more rather than less support, and the company's
immediate earnings goals can only be met from these businesses".
1 I
1!
(Id., p. 76.)
.1
As Jones testified:
"At that point in time we had these two tigers by the tail
other than computers — that is, jet engines and nuclear — and we
^ |! had this host of other ventures that we were trying very hard j
■Q ! to bring on, all of them spun out of this common technology of
*i our electrical industry and electronics industry.
.7 i| I
"We were increasing or had been increasing our debt subs tan- j
.3 ;j tially through this period, so that our debt-to-capital ratio
| has been climbing. And we just said, you know, there is a
.3 | breaking point where we will lose our triple A rating as a j
I corporation if we continue to pile on debt and if we continue
10 j to try to do all these things that we have got on our plate j
right now. I
j "This was the one where we had you might say the least j
H > commitment ... in terms of contractual commitments." (Tr.
8831-32.) j
12
• The Ventures Task Force also had concluded that:
2- I
j "For the first time in our generation, at least, we face the
?5 .j necessity for an allocation of corporate resources which are
; ! -540-
i
i
!
I
2
Ji
i
«•!
si
1
i
i
T\
i
I
s I
i
»!
12 j
13 |
i
i
i
15 i
i
16 I
17 i
:t
1S;|
19 i
20 :{
21 |
22
23
24
25
not adequate to meet all of our readily identifiable needs —
during a period when the company is under special pressure to
demonstrate its ability to grow earnings. The general economic
climate is not favorable; the capital markets are severely
depressed; credit is costly and may be assumed to become pro-
gressively less available;- inflation has forced higher labor
costs on the company following the longest strike in the company's
history." (PX 371-A, pp. 75-76, see also pp. 7, 10; PX 331-A,
P- 7.)
GE could not do everything at once: "[W]e can't afford to
back every horse. We just don't have limitless resources." (R.
Jones, Tr. 8843.)
Disengagement in nuclear energy and jet engines was "not
an available option" because of GE's contractual commitments. (PX
371-A, p. 76.) While nuclear reactors were considered part of GE's
"core business" ("essentially those that dealt with the generation,
transmission, and distribution of the electrical energy in terms of
the equipment to do all those jobs, plus the equipment that would
utilize electrical energy") and jet engines were considered "a spin-
off of the core, but . . . very closely related", General Electric
never "viewed the computer business as being part of its basic
core". (R. Jones, Tr. 8838-41.)*
The Ventures Task Force did consider the possibility of
retrenchment, rather than withdrawal, with such retrenchment taking
the form of moving to a more specialized product line. However, it
was decided that this was not an optimal strategy for General Electric,
* It is instructive that in its "core business", nuclear reactors,
GE had invested for 20 years without making a orofit. (Ingersoll,
Tr. 8288-91.)
-541-
t
r
\
}
J
L
Z
z
£ t
7\
,S !
.9
>0
although it would have required less investment money. (Ingersoll,
Tr. 8144-45, 8150-51, 8188-89; R. Jones, Tr. 8801, 8881-83, 8796-
97.) As Jones testified, this was in part because:
n Vte had sold our equipment to almost every market. We
had not concentrated on banking or manufacturing or retailing.
We had either of our own manufacture or through resale a sub-
stantial and wide ranging line of peripherals, and because our
product offerings over time had been so eclectic, we felt that
if we were to withdraw to the role of a specialist, we would be
in effect abdicating many of the customers and many of the
markets that we had been serving.
"Of course, the result of that would be a substantial
reduction in our overall market opportunities." (Tr. 8797.)
The decision was made to merge a large part of the business
with Honeywell by both companies transferring their business computer
operations to a new Honeywell subsidiary. (DX 555, p. 3.) GE retained
an interest in the new company and, as is detailed below, success-
fully continued its "own independent development of businesses in
the promising areas of process computers, computer time-sharing and
data communications equipment."* ( Id. ) (See pp. 544-46 below.)
As we have seen, the GE decision to disengage was taken in
part because of past GE mistakes and in part because General Electric
!( had other fish to fry.**
;! * GE would have liked to have taken a controlling interest in the
l| new venture but feared the disapproval of the Antitrust Division.
1 (See Ingersoll, Tr. 8242, 8252-5 3; DX 7259, Borch, pp. 13-14.)
J ** Jones testified that he "personally knew of no acts or activities j
22 :| of IBM that would have caused our disengagement", and none were j
| brought to his attention during the activities of the Ventures Task j
2^) Force. (Tr. 8867-68.) !
:! i
25 : ;
i
i
j -542- I
i !
•»l . ■ j
; l !
L
2
3
f
a >
7 |
3 !|
9-
10
11
12. «
i
13 ;
14
15'
IS
17 ! |
13
Could General Electric have succeeded had it remained?
Bloch testified that he thought GE had the resources necessary to
become "a clear No. 2 in the supply of computer systems used for
business data processing purposes". He ,r felt at the time that it
■was a mistake" not to implement the APL. (Tr. 7811.) Withington,
who, unlike Bloch, was not personally involved, testified that GE
had sufficient assets to be successful if it chose to invest them in
the general purpose computer systems business. (Tr. 56732.)
Indeed, General Electric did not have much trouble raising
money in the years after 1970. From 1970 to 1974, it raised approxi-
mately $625 million by long-term debt. (DX 553.) It chose to
invest those funds elsewhere.
i. Did GE Lose Money? It is questionable whether GE lost
much, if any, money in the course of its computer operations or, at
least, whether it would have lost money thenceforward had it not
sold part of its operations to Honeywell.
I A report- by Peat, Marwick & Mitchell, General Electric 's
utside auditors, showed that GE ' s domestic business computer opera-
tions lost approximately $163 million in the period 1957 through
19 i
■{September 1970. (PX 380; see Ingersoll, Tr. 8353-56.) Jones testified
20.1
jthat this accorded with his recollection. (Tr. 3756.) Ingersoll
21 =1 .
testified that this figure included allocation of corporate overhead
22 '">
expense to GE ' s domestic business computer operation, involving
23 |
expenses which continued after the transfer to Honeywell and which
v-\
yere not directly incurred in the operation of GE ' s domestic computer
\\
-543-
i
3 i
operations. (Such allocation amounted to 10% to 15%.) (Ingersoll, Tr.
8359-60 , 8365-66.)
Further, Ingersoll testified that PX 380 "reflects the
cost of developing the equipment incurred by the Domestic Business
Computer Operations that were subsequently transferred to Honeywell—
the development/ that is, through the date of the transfer". (Tr.
j\\ 8377-78.)
3
3
a
i
2
.5
Hence, in order to evaluate whether GE would have ultimately
made losses had it not sold part of its operations to Honeywell, one
must consider the profit stream which would have resulted from
those sunk expenses. As Ingersoll testified, "because of the rela-
tively long life of computer equipment, ... in order to properly
evaluate the total results one should look toward the full life of
the equipment" . He agreed that it is "generally true that in the
development, manufacture and marketing of new lines of computer
• equipment that losses are sustained in the early years and profits
;| derived in later years". (Tr. 8199-200.)
" !|
!| According to GE's proxy statement, that portion of GE's
• s J
:| computer operations which was sold to Honeywell had an after-tax
j profit for 1969. (DX 554, p. 9; see Ingersoll, Tr. 8329-30.).
10 \
I Further, in December 1969, GE's Information Systems Group estimated •
11 : !
j that the net income from its current computer product lines for the
22 '}
.] years 1970 to 1975 would be $173 million (this estimate also included
2 I
a positive impact of APL on current products). (PX 362, p. 10; see
24. 1
j Ingersoll, Tr. 8378-79.). The Ventures Task Force in April 1970
5S I
• -544-
i
i
.1
.j
I
2 !
3J
3
a
estimated that the current product line would bring in $821 million
in revenue and $164 million in net income in the years following
1969, regardless of APL (it was evaluating the business from the
point of view of a prospective buyer) . (PX 331-A, p. 32; see
Ingersoll, Tr. 8376-77.)
~ lj The terms of the sale, which was announced on May 20, 1970
jt (PX 323 (DX 14502)), were as follows. The two companies formed Honeywell
fl r Information Systems (HIS) and GE received an 18-1/2 percent interest
3 !
in it. In addition, GE received 1.5 million shares of Honeywell
common stock and $110 million of Honeywell subordinated notes (later
converted to additional shares of Honeywell common). (DX 555, pp.
22-24; see Ingersoll, Tr. 3393-96; DX 14073, p. 32.) At the time GE
recorded a profit of $1.7 million on the transaction. That amount
was quite conservative, GE having undervalued its minority interest
and the value of. its Honeywell stock. (DX 555, pp. 26, 31.) Ingersol]
testified that the market value of the 1.5 million shares of Honeywell
stock received by GE was "in the neighborhood of $130 million".
(Tr. 8388-89.) In addition to that and the $110 million in notes,
Ingersoll testified that GE valued its minority interest in HIS at
"approximately $32 million", "substantially less than the net book
value of that minority interest as determined by Honeywell" which
valued it at "at least a hundred million dollars". (Tr. 8393-96.)
In fact, the 1,500,000 shares of Honeywell stock had an average
market value of about $120 million as of the last quarter of 1970,
based upon the average between the high and low prices at which it
II
12.
13
14
15
IS
17
IS
19
2a
21
22
22
24
25
-545-
^ traded during that period. (DX 555, p. 31; DX 14064, p. 164.) In
2 1971, GE received 1,025,432 shares of Honeywell stock in exchange
3 I for the $110 million in notes. (DX 14073, p. 32.) Those shares had
a ; an average market value of about $113.2 million in 1971. (DX 14-130,
5 • p. 164? DX 14131, p. 172? DX 14132 r p. 172; DX 14133, p. 173.)* In
g ; 1976, GE exchanged about one-third of its interest in HIS for 800,000
j shares of Honeywell stock and, in 1977, the remaining two-thirds of
g another 1,400,000 shares. (DX 13980, p. 40; DX 13981, p. 40.) The
g ' Honeywell stock received in 1976 had an average market value in that
1Q ( year of about $35.6 million (DX 14062, pp. 2, 4, 6, 8); the stock
11
received in 1977 had an average market value in that year of about
12 $68.5 million. (DX 14063 r pp. 2, 4, 6, 8.) GE sold the Honeywell
13 stock over the years, disposing of the last in 1978. (DX 13981, p.
14 40; DX 13887, p. 40.)
15 Taking all these things into account, even allowing for
lg the difference in timing and inflation between the expenditures made
27 . in the early sixties and the returns received from the Honeywell
lg , sale, GE appears to have been a net gainer in the computer industry.
19
20
21
22
23
24 * As used herein, the average price of Honeywell stock in a
given year is the average of the high and low prices at which it
25 traded during that year.
-546-
L || 42. RCA. As we have seen, the 1950s were a stagnant
i '
Z \ period for RCA' s computer business. During the early 1960s RCA
^ : | experienced several problems that continued to retard its growth in
4* [\ computers, in particular the failure of the 601 and its on again-off
again peripherals development. (See pp, 196-207 above.) Toward the
5 f middle of 1963, RCA had stopped marketing the failed 601; the 501
jl
7 i| was "starting to decline" and RCA was marketing only one computer
3
io :
13
14.
model, the 301. (McCollister, Tr.. 9622.)* RCA then announced its
3301 computer system.
a# RCA 3301. The RCA 3301 was announced on August 20,
1963. (DX 580, p. 1.) The 3301 was an "interim product", designed
and marketed by RCA "to round out our overall product program . . .
[by] tak[ing] the place of the 601". (McCollister, Tr. 9247, 9622.)
The 3301
"was not a new design. It wasn't intended to be the
foundation of a future line of products; rather, it
T c ': was a product that we could develop relatively quickly,
* ;; at relatively low engineering expense, [**] that would
17 |
18 i
TCI
• -j * E. S. McCollister joined RCA in 1961 as Vice President of
II Marketing for the computer division. He held that or a similar
- a .| position until December 1971 when he left RCA. He then joined the
j Burroughs Corporation in January 1972. (McCollister, Tr. 9161-62.)
j ** McCollister recalled that the engineering development cost
22- -j for the 3301 was "in the order of about $'2 million for the
:j processor and for the associated control units". (McCollister,
22 Tr. 9623.)
2*1
i
25 1
-547-
give us an additional offering to take the place of the
601, and that in a sense would give us time to get on
with a complete new product program in the longer
range future." (MeCollister, Tr. 9622-23.)
RCA described the 3301 as an ,r all purpose computer" that
"features advanced communications devices and arithmetic circuitry
to make it equally powerful for scientific equation solving, super
fast business data processing, instantaneous (real-time) management
control, and high-speed data communications". (DX 580, p. 1; see
also Beard, Tr- 8994-95.)* In addition, the 3301 could perform
applications "which were in other circumstances performed by multiple
installations of special purpose computers". (Beard, Tr. 8994-
95.)**
While a more successful product than the 601, the 3301' s
success was limited for two reasons : poor peripherals and its
"eclipse" by the announcement of the Spectra 70 series less than a
year and a half later. (Beard, Tr. 8458-60, 10276, 10307.)
1 : RCA had been beset by problems with the peripherals used
M
r
X
on its 501 and 301 systems. (See above, pp. 195, 201.) The peripherals
on the 3301 "prevented the computer system from achieving its full
* : j throughput capabilities". (Beard, Tr. 10276.) Although RCA had
j
^ ;j resumed manufacturing its own peripherals in 1962 (see above, p.
T I
~ j * The 3301 was not as versatile, however, as System/360. (Beard,
- ! Tr. 10266.)
I
A j ** A. D. Beard, Chief Engineer of the RCA computer division from
~ j 1962 until 1970 (Beard, Tr. 3447-51) , used the term "special purpose
- | computer" to mean "real time" computers such as SAGE (Tr. 8995-97),
~ | "communication equipments", and "small and medium sized scientific
j computers". (Beard, Tr. 8996.)
i
1
■I
': -548-
I
2
3 j, existing on the 301 or which could be made available from outside
4-
f
7 I
3
9-
10
11
12
13
14-
15
202) , time and continuity of effort had been lost and RCA was
largely "constrained to live with the peripherals that were then
suppliers". (Beard, Tr. 9004, 1.0307; see also McCollister, Tr.
9622-23.)*
RCA also experienced problems with a peripheral unit of its
own manufacture, the RACE mass storage unit. The RACE unit was a
storage device that used magnetic cards . ** Those cards
"had to be extracted from a magazine, put in a channel that
carried it to a revolving drum, held on the drum while it
rotated past a reading head, where the information was read or
reported, and then the card had to be returned to the magazine
from whence it came.
"And this was a very, very complex mechanism and a very
difficult technical task." (McCollister, Tr. 9657.)
RACE was designed to provide random access storage for the
3301. Compared to IBM's 2311 disk drive,/ announced by IBM with
System/360 in April 1964, RACE was "much smaller" in terms of storage,
15 :! but "considerably faster" in terms of access time. Thus, for the
*7 !| application mix of some users, RACE, when operating properly, would
-S i| be superior to a disk drive and under other circumstances , the disk
19 j
11
20 '| * According to McCollister, "In the 3301 [RCA] used a card reader
I from Uptime, we used the ICL, or ICT I believe it was at the time, a
2- ;j hundred-card-a-minute punch, we used the Anelex printer, of which we
i bought the complete printer sub-system from Anelex." (Tr. 9622-23.)
22 '■'>
•[
ij ** The RACE unit came in two models, the 3488 to be used with the
23 | 3301 processor and the 568 for the later Spectra series. (McCollister,
ITr. 9656-57.)
24 'I
;! / RCA did not have a disk product of its own to offer with the
25 ;j 33-01. (Beard, Tr. 9046.)
-549-
drive was superior. (Beard, Tr. 9046-47.)
The major problem with RACE was that it was not reliable.
Thus, Wlthington classified RACE as "a major product failure" (Tr.
56511) , and McCol lister explained why:
"[Tlhe cards wore out . » . the cards were damaged in transit
. . . sometimes there was a failure to select the proper card
... it was a tedious process to replace a card in the file
when it was beginning to wear out and, indeed, to detect when
it was beginning to wear out." (Tr. 9658.)
Moreover, even when operational the RACE unit was "unable to meet the
speed of accessibility that had originally been specified in the pro-
duct" . (McCollister, Tr. 9658.)
Second, the success of the 3301 was limited by RCA's intro-
duction of a new series only a little more than a year after the 3301
was announced. RCA's announcement of its Spectra 70 series in
December 1964 "eclipsed" and "superseded" the 3301. (Beard, Tr.
8458-59; PX 4830, p. 25.) Thus, potential customers of the 3 301 were
encouraged to obtain Spectra 70 series. For example, RCA provided
emulation of the 301 on the Spectra 70 but not on the 3301. (PX
4830, p. 25.) Another sign to RCA's users that the 3301 was a dead-
end machine was RCA's failure to provide a "growth machine" for users
3 |
\ of the 3301.* (Beard, Tr. 9986-87.) Because RCA provided no emulation !
3 i i
i from the 3301 to Spectra 70 or to any other system, 3301 users had ;
3 ' I
2. ! l i
3 s * a "growth machine" is one "which allows the programs to be moved
i from the predecessor machine to the growth machine [with] a minimum
.4 ! of re-programming effort". (Beard, Tr. 9986-87.)
■«
-550-
nowhere to go in the RCA line without converting their programs.
2 j (Beard, Tr. 8458-59, 10235.)
b. The Spectra 70 Series. IBM announced its System/360
4-j on April 7 , 1964. Beginning "shortly" after the announcement, RCA
formulated the "design specifications" for its Spectra series. Those
5 ';! specifications were done "in [a] preliminary fashion" around July or
August of 1964. (McCollister, Tr. 9624.)
The Spectra 70 series eventually comprised eight models —
the 70/15, 70/25, 70/35, 70/45, 70/46, 70/55, 70/60 and 70/61. The
sizes of the processors increased in numerical order, and the 70/46
and 70/61 were intended to offer time-sharing capabilities. In
December 1964 RCA announced the 70/15, 70/25, 70/45 and 70/55.
(Beard, Tr. 8483-85; McCollister, Tr. 9635-36; DX 669, p. 11.) The
70/35 was announced in September 1965 (DX 670, p. 16); the 70/46 was
announced in 1967 (PX 338, p. 22); the 70/60 and 70/61 were not
announced until 1969 (DX 674, pp. 8-9) .
la |i
j No prototype of any of the systems was in existence at the
I time of the announcement. A prototype of the first machine was not
IS :|
j built until the middle of 1965 at which time prototypes of most of
19 i
ir the control units were also built. (McCollister, Tr. 9635-36.)*
7 s
a ;
9 j
ia j
ii j
12 j
^ \
L4 ;
?T
! * During 1965 through 1966 several new peripherals were also
— :• designed. (McCollister, Tr. 9635-36.) During 1965 and 1966 RCA's
i total engineering budget for software and hardware was approximately
-^ j 315 million annually, out of which came engineering expenditures
mjk -j for the support of older products as well. (McCollister, Tr. 9634-35.)
-" j Compare the status of RCA's Spectra at announcement with that of
I IBM's System/360: 3y the time of the IBM 360 announcement there
- 3 j were prototypes of all models of the processors. (3rooks, Tr.
•j
■! -551-
Deliveries of the "small systems" began in 1965 and the "larger
systems" in 1966. (Beard, Tr. 8460.)
Four aspects of the Spectra 70 series are particularly
important: its attempt at compatibility with System/360, its ability
to perform commercial and scientific applications, the problems RCA
encountered and the success of the Spectra 70 series.
(i) Compatibility with System/360 . In a decision that
affected both the Spectra 70 and its succeeding RCA Series, RCA
decided to make its Spectra 70 series compatible, that is, able to
use the same application programs -with little or no modification,
with IBM's 360 systems.
By making its Spectra 70 compatible with IBM's System/360
RCA hoped to be able to persuade substantial numbers of 360 users to
move to the Spectra series. (Beard, Tr. 8461-63; McCollister, Tr.
! | 9269-70; Rooney,* Tr. 12117.) In particular, RCA expected to target
<n
7 ! 22695-6; Hughes, Tr. 33995.) Most of the processors and some of the
! j peripheral equipment were in the early stages of product test
3 .J (McCarter, Tr. 88382-83; JX 38, 1! 19); all, or almost all, the
'I memories had undergone technical evaluation testing (Brooks, Tr.
3 \ 22699) ; microprogramming and multiprogramming had been tested on
; j the Model 40 (McCarter, Tr. 88382-83) ; four estimating, forecasting
£ I and pricing cycles had been completed; and the "componentry,
I systems and product testing program already completed . . . [was] more
LI '■ extensive than the entire program ever [previously] undertaken for a
3 system". (DX 1172, p. 2; see also DX 1165.)
j * J. W. Rooney joined RCA as Vice President for Marketing Operations
23 in 1969. He became the Vice President of Marketing for the Computer
I Systems Division in 1970 and was President of the Division from 1971
Ml until he left to go to Itel in 1972. (Rooney, Tr. 11687-88.)
j -552-
L
2
3
a
7
3
9
LO
11
12 - ;
13
14-
15
IS ;
;i
n \
its marketing efforts at those 360 users who wanted to obtain
larger or more functional equipment. (Beard, Tr. 85 26-28.*) To
those users RCA wanted to offer better price/performance on its
Spectra 70 equipment than IBM did on its 360 equipment and thus
persuade the user to acquire Spectra 70 equipment. (Beard, Tr.
10103; Rooney, Tr. 12117.)
Thus, during this period RCA attempted to offer a price/
performance advantage of between 15% and 20% over IBM's systems.
(Beard, Tr. 8493-94, 10095; Rooney, Tr. 11826; see also Wright,**
Tr. 13083.) RCA's pricing methodology was the direct result of
RCA's strategy to be compatible with IBM and to go after IBM users.
* RCA believed that compatibility had become a more useful
marketing tool after introduction of 360. Prior to 360 RCA had a
"sales opportunity" whenever a customer wanted to go to another
computer because:
"As the computer industry evolved, ordinarily even
in moving within the line of one manufacturer, within
;| the line of IBM, there was conversion that was necessary.
17 ;{ And since a customer was facing ... a conversion in the
: case of IBM, we could argue it won't be any more difficult
13, to convert to RCA." (McCollister, Tr. 9273-77.)
IS ; i Since users of 360 would be able to move up to a larger IBM
"I 360 computer "relatively easily", it would be to RCA's
20 j advantage "to make it almost as easy as we could for the cus-
tomer to move to the RCA product line as to move up within the
IBM product line". (McCollister, Tr. 9273-75.)
** V. 0. Wright joined RCA as Vice President and Regional Manager
of Federal Government Marketing in the Computer Division in 1970,
became head of Systems Development in 1971 and left in 1972 to go to
! Amdahl. (Wright, Tr. 12785.)"
24 !
i
-553-
"[I]f you are going head to head with a competitor,
such as IBM, and you essentially are going to offer the
same function and if the competitor is in a stronger
position in the marketplace than you are, you would
offer a price advantage to move your product." (Rooney r
Tr. 12415, see also Tr. 12414, 12420-21.)
The pricing methodology was based upon two assumptions:
(1) RCA assumed that in many cases Spectra would be
offered to displace existing IBM computers and some
inducement would have to be offered to persuade the
IBM user to go to the trouble of replacing his existing
IBM computers and install Spectras (Beard, Tr. 10103;
Wright, 13083-84)*; and
(2) Spectra was delivered one to two years after
IBM's System/360, and customers expected a new offering
to have a price/performance advantage over older computers .
(Beard, Tr. 10103-05.)
If RCA had not adopted the compatibility strategy, other pricing
strategies would have been available:
' i "If you did not have compatibility, you would be
going with your own product line, which would have its
3 ; t own unique characteristics and functional capabilities.
s\
a |
:i * This "was needed in order to compete with IBM and in
— j order to obtain business from people who were currently using IBM
'j systems, to displace IBM systems. If a customer has a fairly
£ ;| substantial investment in software, he has to have some reason and
j some motivation to move to another vendor, and we felt that that was
S | the margin of motivation that was needed to get them to move."
| (Wright, Tr. 13083-84.)
i
15
e ■
:i
-554-
L
2 !
i
i
5 : |
I
& i
7
3
10
II
12
13
14-
15
la
17
13
•j ". . . [Tjhere would be some possible configurations
2Q ] whereby you would not meet your price/performance goal
:| of 15 to 20 percent; you would only meet them on the
2£ j specific points that you had evaluated.
22 •; "There would be some cases where you would
j exceed that price/performance advantage. There would
~- ;j be other cases where perhaps the advantage would go
24 1
I
•(
] -555-
"If you felt that you were matched in those unique
capabilities and functions by IBM, you would probably
price your products for comparable systems in the same
range but not necessarily under IBM.
"If you felt you. had a unique product, then you
would not be guided by IBM's pricing policies. You
would then price based upon your analysis of what the
market would bear for that functional capability."
(Rooney, Tr. 12421.)
RCA employed the technique of "straddling", placing its
machines in terms of performance "approximately midway between a
pair of IBM machines". (Beard, Tr. 10106-07, 10113, 10121-22, see
also Tr. 10097-99.)
"Generally what was done was to pick what
appeared to be the most commonly used configurations
of equipments : so many tape drives for a small
system, so many tape drives for a large system, and
so forth, to pick several of what were considered to
be representative points around which you would expect
a large number of customers to cluster.
"Based on that simplified set of system con-
siderations — that is, not taking all the possible
combinations into account, but some of the most probable
ones— evaluations were made on a system basis, that is,
not only the. speed capabilities of the main processor
but also what range of peripheral speeds you would put
on each of these system configurations , determine what
the relative overall performance advantage or dis-
advantage was, and set the prices accordingly.
the other way." (Beard, Tr. 10092-93; see also Rooney,
Tr. 12129-31.)*
As discussed at greater length below, RCA's attempt to offer
better value to the customer than System/360 failed even though it
may have been able to announce that it provided more throughput per
dollar (better price/performance) than some of the 360 systems with
which it competed.
RCA also believed that the compatibility strategy would
have a good chance for success because it could take advantage of the
situation where IBM users were leasing from IBM on a short-term
basis, which leases could readily be terminated and RCA equipment
substituted. (Beard, Tr. 10073; see also Rooney, Tr. 12126-27.)
A number of arguments against the compatibility strategy
were raised at RCA:
First, if IBM customers could switch easily to RCA machines,
then RCA customers could also switch easily to IBM machines. Beard
5 'I
• called this a "two edged sword". However, RCA felt that it "had more
7 ; ! . .
y * to gam . . . than to lose" because IBM had many more existing cus-
^ ;, tomers than did RCA. (Beard, Tr. 8519-20; see also Rooney, Tr.
.i
9 I 11857.)
LI |
„:{ * Thus, the Spectra 45 was placed between the System/360 Model 40
: j and the Model 50, and the Spectra enjoyed a price/ performance
y I advantage over the 360 Model 40. However, the performance of the
^ [ 360/50 was sup«rior to the Spectra 45. The same was true with
1A J respect to the comparison between the Spectra 55 and the System/360
— | Models 50 and 65. (3eard, Tr. 10106-07, see also Tr. 10097-99.)
J The Spectra 4 5 and 55 were bid against the IBM System/360 Models 50
^ j and 65. (Beard, Tr. 10113, 10121-22.)
i
■1
\
\ -556-
■V
1
i;
i
, i
u
£ i
L
2
3.
4.
f
5
7
3
9
10
II
12
12
14*
15
IS
17
IS
Second, the similarity between Spectra and 360 "sharpened
the comparisons" between RCA and IBM., making it "easier for the cus-
tomer to analyze and quantify the differences" and putting "RCA in a
position where its products could easily be criticized versus what
IBM was offering. ... If there were any deficiencies on RCA's
part they would probably stand out as weaknesses." (Beard, Tr. 8526.)
Third, RCA could have chosen "the most natural alternative
. . . an extension of the 301, 3301 systems". This would have pro-
vided two advantages to RCA: it would have given it "a certain
advantage" in marketing to the existing 3301 user base, because of the
"software investment that [the users] had made in those machines".
(Beard, Tr. 8524-26.)* And it might have enabled RCA to provide a
"superior architecture to what IBM had chosen".** (Beard, Tr.
8524-26.)
RCA considered this important decision only briefly.
McCollister testified that "because of the press of time in this case,
;( * As noted above, RCA chose to forego this "advantage" and did
19 .i not provide any emulation on the Spectra 70 series for the 3 301
t user.
I ** According to Beard this possibilitv did not seem to be very
21 ! likely:
•4
22 :! "However, I think most of us felt that it really didn't
| make that much difference to the customer what particular
23 | machine instructions were made available; that the 360
I set was a complete set, it included most of the things
2- 1 ' we had thought of and perhaps some that we had not thought
■! of; there were some things that were missing, but these
25 I were secondary in our minds." (Tr. 8524-26.)
-557-
I am not even sure that there was a formal product proposal". (Tr.
9630-32.) The compatibility arguments prevailed, and two or three
weeks after the announcement of System/360, RCA decided to make its
Spectra series "as compatible with the 360 as the circumstances per-
mitted". (McCollister, Tr. 9273.)
With the compatibility approach that RCA chose, its Spectra
series had the same instruction set, instruction format and word
length as 360 » However, "in terms of the engineering implementation
of this architecture, it was quite different between RCA and IBM.
.... If you took these machines apart, they were totally different
machines . . . RCA used a completely different set of components . "
(McCollister, Tr. 9644-45.)
(ii) Commercial and Scientific Ability of Spectra 70 Series.
As discussed above (see pp. 290-96), IBM's System/360 was aimed at
all users regardless of application. Initially, RCA planned to market
the Spectra systems for "commercial as distinct from scientific
purposes . . . it' was a stated strategy to all of our marketing people
that we were selling to the business environment and precisely said
that we did not have a computer to compete in the scientific arena".
:| -558-
i
;l
L
2
3
4-
Series" was a "versatility for handling data processing, real time,
and scientific applications from the small user to the very large".
(DX 617, p. 2; see Beard, Tr. 9099-100.) Beard testified that
the "primary reason" for making that a "salient feature of the design
philosophy of the Spectra 70" was:
"We felt that as the customer world became more sophis-
ticated that there would be a consolidation in the
computer type operations of more than one type of
function and therefore this versatility, which allowed
for engineering and scientific type problems, communi-
cations problems, data processing, batch problems,
information control systems . . . could be merged into
one computer complex.
"It may be a relatively small computer, if the
operation is a small one. It could be a very large
computer, if the operation had a large volume of data
to be processed. There would be configurations where
reliability was extremely important and it would be
necessary to have multiple processors in order to have
redundancy in the system.
7;
a;
* j
!
10 :
i
t
11 :
12. |
13 \
"It would be an advantage, from what we saw in the
5- ;| field emerging, for the hardware as well as the software
to accommodate these various functions in one system
t - ii as opposed to having distinct unique systems for each
* & of those functions." (Tr. 9100-01.)
17
IS
Very soon after the initial delivery of the Spectra, the
consolidation of the various types of functions which was anticipated
•a*
±* :j in "the design philosophy of the Spectra 70" had come to pass with
;(
2G ;j "some of the more advanced customers . . . ready for these types of
.1
-- ; ! systems in the latter half of the sixties, and certainly that trend
22 1 has continued into the seventies". (Beard, Tr. 9101-02.) By 1970
23 j RCA was advertising the versatility of the Spectra 70:
r
I
2* i "The emergence of third generation equipment with
j increased speed and storage capacity has brought us
25 j to the realization that scientific applications are
-559-
L
Z
3.
3
a
7
8
9-
LO
11
12
13.
14-
15
16
17
13
19
20
21
within reach of almost every computer user. In the
past these applications were confined to the big
and expensive machines.
"For all your, data problems — from simple account-
ing to management science programs — Spectra 70 offers
a complete systems approach.
"Linear programming . . . statistical analysis
. .. . simulation » . . automatic machine tool control
. . . all are key elements of management science
operations » Spectra 70 handles these applications
and your normal data processing at the same time . "
(DX 619, pp. 1, -39, emphasis in original.)*
To assist its customers RCA offered its "Systems Scientific Services":
"Systems Scientific Services provide a broad range
of support, to RCA customers in software areas.
"By supplying generalized scientific, statistical,
simulation, and mathematical software, Systems Scientific
Services assist the user in achieving efficient use of
his system . . . right at the start.
"Also available are scientifically oriented pro-
gramming systems." ( Id. , p. B»)
(iii) Problems with Spectra 70 Series. The Spectra 70 series
suffered from various problems that hurt its performance in the
marketplace. Much of the equipment of the Spectra 70 series suffered
from reliability problems which users took into account in choosing
RCA also explained the new field of management science:
• : i "Along with the advancement in equipment, the
22 :j technology and management science has made significant
,.| advances to the point that it is an integral part of
23 j modern, efficient, organizational management.
24 I "Management science has a broad definition that
1 includes mathematical, statistical, and operation
25 ; | research techniques that aid in effective decision
making on the part of management." (DX 619, p. 1.)
-560-
z '
i
3
f
7
8
9
LO
11
11
n.
15
16
between RCA and IBM computers. (Rooney, Tr. 12190-91.) Rooney com-
plained about RCA's equipment as late as June 1970:
"RCA equipment apparently requires larger amounts of
dedicated preventive maintenance time than that of our
main competitor, IBM. Customers that have both our equip-
ment and IBM equipment are aware of this , and this works
to our detriment in the marketplace." (DX 621; see also
Rooney, Tr. 12186, 12202-03.)
Moreover, according to Rooney, RCA's equipment was
"apparently more sensitive to environmental fluctuations
than that of competition, particularly to IBM. This makes
our customers somewhat sensitive to the differences between
our maintenance policies and theirs. I am told, for
example, the 360/30 's can be left without any maintenance
whatsoever for weeks on end. Yet, most of our systems
require that we take the system from the customer for
periods of time every day." (DX 621; see also Rooney,
Tr. 12145-48.)
For example, RCA's disks were "more sensitive to air conditioning"
than those of IBM, "so, if you did not have the adequate amount of
air conditioning, that could lead to the need for more preventive
maintenance". (Rooney, Tr. 12197-201.)
RCA suffered problems during the installation of the
*7 ,| Spectras. In that regard Rooney testified that:
-S ;j "RCA equipment was more difficult to install because
:| of certain environmental factors. I remember the RCA
-- \ equipment required more air-conditioning and power and
3 I remember a problem of size, physical size of the units
-- ,| being involved, in terms of: if we replaced IBM, certain
i of our units would require more physical floor size than
21;! IBM equipment." (Rooney, Tr. 12175-76.)
•i
22 '\ These problems made it harder for RCA than for IBM to install its
23 {equipment. (Rooney, Tr. 12204-05; DX 620.) It was reported to
i
24. 1 Rooney in 1970 that:
-561-
"In the area of installation, the RCA-IBM comparison
is not restricted to just power and air-conditioning
requirements. The problems are more profound, and
bear directly on the equipment designed.
"The installation of RCA data processing equipment
has historically been more difficult and more time, con-
suming than. that of our competition, particularly IBM's.
Since the RCA marketing strategy is to sell to the IBM
replacement market, the installation of RCA equipment
is constantly being compared against IBM in an unfavor-
able light." (DX 620; see Rooney, Tr. 12205-06.)
During 1968 a portion of the marketing force was diverted
from seeking new business to coping with problems of installation.
At that time the marketing force was
"very, very heavily occupied in working with existing
customers on the installation of equipments which has
[sic] been ordered at an earlier time. . . . [D]uring
the year 1968 about 75 percent of [the time of] the
marketing organization * . . was devoted to working
with existing customers as opposed to seeking new
business. . . . And this made very heavy demands upon
^ ~| the time and capabilities of our field marketing
organization, and this impacted to some degree our
ability to get new orders." (McCollister, Tr. 9647-50,
9653.)
7
\
i
A
RCA also found that its marketing force had to take time out from
their normal selling efforts to deal with "[t]he problems of training
* ij customers in the programming of the equipment, in working with the
'I
- :j customer in the installation of the equipment and the conversion of
;{ !
- | his system of processing work to this new method". (McCollister, j
- i Tr. 9649-50.)*
3 -t * The amount of effort expended by the marketing force on customer !
j training was related to the fact that "the Spectra 70 equipment was
& I new to the user" . There was a demand for the services of the mar-
:| keting organization to deal with problems in systems programming for j
15 j Spectra because there were new programming products and, as McCollister :
-562-
I
2
Z
4
5
S i
7
3
9
10
11 !
12
13
In addition to the problems that pervaded the entire Spectra
line, RCA experienced problems which were uniquely associated with
particular models. Those problems caused the Spectra product line to
vary greatly in its degree of success .
The Spectra 70/15 and 70/25. First deliveries of the
smallest computers in the Spectra series, Spectra 15 and 25, were made
toward the end of 1965. (McCollister, Tr. 9640.) The 70/15 and 70/25
had less function than the rest of the Spectra line. This permitted
them to be brought out earlier (Beard, Tr. 8460) :
"They did not use integrated circuits . . . [i]t
was a means of protecting us against any risks that
there might be in the use of integrated circuits in
the larger systems .
"Secondly, we felt that there might be some cus-
tomers who are interested in a system that did not
have a complete instruction set and was simpler to
14, ji operate." (McCollister, Tr. 9719.)
The 70/15 and 70/25 were "relatively poor competitors"
(Beard, Tr. 10110) and thus not very successful. (McCollister, Tr.
9642.) The lack of a complete instruction set — one of the reasons why
the 70/15 and 70/25 were introduced — and the limited capability of the
systems were two of the liabilities of the 70/15 and 70/25:
"It turned out that most customers wanted the
systems which had the more complete capabilities, and
also the 70/15 and 70/25 did not have the communica-
tions capabilities that the larger systems had and they
did not have the programming language capabilities that
15
16 (
17 \
IS j
i
19 ;
20 :
21
22
23 •,
,\ put it, "the experience of the industry in general is that there
Z-" ;) is always work to be done on new programming products". (Tr.
9651-52.)
25
-563-
r
\
}
1
I i
L
a
4-
the larger systems had". (McCollister, Tr. 9719.)
In addition, "there was no COBOL capability provided at all" on the
70/15 and 70/25 (McCollister, Tr. 9730-31) , which was anomalous
since the 70/15 and 70/25 processors "in general left out the
scientific type of instructions, and concentrated primarily on the
data processing instructions". (Beard, Tr. 9071-73.)
In November of 1968 the product planning organization of
the RCA computer division* made a similar observation:
"[T]he Spectra 70/15 and 70/25, are basically sound
processors, however, minimal software, no communication
facilities , no random access hardware or. software
facilities and the lack of slow speed/low cost card and
print devices were the prime reasons for poor competi-
tive position." (PX 127, p. 77.)
The 70/15s and 70/25s were also hurt by RCA's absence of
"marketing emphasis". (McCollister, Tr. 9729.) McCollister testified
that the competitive position of these two systems "really wasn't that
important to the RCA computer division". They were "insurance
* i policies" using "existing technology that we could bring to market,
.
7 I deliver to customers before we could deliver the larger systems".
•t
i
3 ;j (McCollister, Tr. 9740-41.) As a result RCA put little effort into
« 1
5 .} marketing the 70/15s and 70/25s:
a!
>7 I
~ j ' * During the period 1964-1972 the RCA computer division had three
*5 :! different names: from 1964 to 1968 it was called the "EDP Division";
"" i from 1968 to 1970 it was called the "Information Systems Division";
and after 1970 it was called the "Computer Systems Division". Here
it will generally be referred to as the "computer division" or
"Computer Systems Division" unless appearing in a quotation.
La
je
-564-
L !
2
3 i
5" i
»
i
T I
3
9-
a
12
13
14*
15"
^ ,
* In distinct contrast to the RCA lassitude with the low end of
17 I its line, IBM constantly attempted to grow the market with its low
; end computers. Thus, the 1401 (announced in 1959) and the 360/20
IS | (announced in 1964) were the largest-selling IBM systems of their
j time. (See above, pp. 141-42, 399.)
19 i
j ** The Spectra 45 was first installed in July or August 1966 and
20 ! the Spectra 55 first installed toward the end of that year. Deliveries
i in quantity of the 70/45 began in 1967. (McCollister, Tr. 9640-42.)
21 :■!
i /Beard wrote in 1965 (and testified as to the accuracy of his
22 1 statement) that "the Spectra 70/45 is a medium-scale processor with
a high performance capability for business, scientific, communications,
and real-time applications", giving airline reservations or brokerage
| quotations as forms of real-time applications. (Beard, Tr. 9080-81;
24. 1 DX 617, p. 7.)
I
I
i
•! -565-
"These were relatively low cost, low margin systems,
and when we had a finite amount of marketing field
manpower, it made better economic sense for us to
concentrate on the larger systems , where we had larger
unit sales value J* (McCollister, Tr. 9724-25.)*
The shipments of 70/15s and 70/25s turned out to "trivial".
(McCollister, Tr. 11355.) And RCA produced the 70/15s and 70/25s
"only during part of the total life cycle of the Spectra 70 family".
( Id. )
The Spectra 70/35, 70/45 and 70/55. The Spectra 70/35,
70/45 and. 70/55 were larger processors than the 70/15 and 70/25.
Deliveries of those systems were about "fifteen months or so behind
IBM".** (McCollister, Tr. 9646.) The 70/45 turned out to be the
"most successful" of RCA's Spectra series./ (McCollister, Tr.
9665.) The 70/55 was less successful than the 70/45. It suffered
from several problems:
(1) The 70/55 "had serious memory problems. . .
We would get repeated errors in memory due to tech-
nical failure in the memory itself and this would
bring the system down. [We] had a great deal of
difficulty in maintaining the gear and keeping it
up." (Rooney, Tr. 12139.) In fact, "there was some
exchange of memories . . . some early number of the
first machines had to have their memories replaced."
(Beard, Tr. 10111-12.)
(2) RCA experienced "manufacturing problems
with the [70/] 55s, which gave us an unusual amount of
field maintenance attention during the first year" .
(Beard, Tr. 10112.)
(3) The 70/55 was difficult to install and
relocate. (DX 620, pp. 1, 3-4.)
(4) The 70/55 came out approximately a year
after the 70/45 and tended to be "eclipsed" by the
70/60 and 70/61 which RGA brought out shortly there-
after. (Beard, Tr. 10109-10.)
2Q '■[ (5) The 70/55 was hurt because it did not offer
2i ;j any emulation capability. Notwithstanding the fact
22 ' : j that emulation capability was important to the success
23 ij of the 70/45 system, the Spectra 55 did not emulate
24 !J anything. (Beard, Tr. 10109, 10233.)
25 ,i
-566-
L
2
3
4*
5
a
7
3
9
10
12
13
14-
15
The result of these failures of the 70/55 had a "dampening
effect on the [RCA] sales force" and led to customer cancellations.
(Beard, Tr. 10111-12; see McCollister, Tr. 11216.)
While the 70/35 provided for emulation of the 301, that
emulation "did not work successfully because the Spectra 35 was priced
at such a high price that it was not a logical move for the 301 user
to move up to the Spectra 35 system. . . . [Thus,] 301 users did not
move up to the Spectra 3.5". (Rooney, Tr. 12137-39.)
The Spectra 70/46, 70/60 and 70/61. RCA's 1969 Annual
Report described the 70/60 as a
" [l]arge-scale . . . batch processor, which is designed
to handle retail credit and reservation systems, auto-
mate production control, and service government and
industry data banks.'* (DX 674, p. 8.)
The 70/46 and 70/61 were time-sharing systems. RCA began its work on
time sharing during 1967 by attempting with the 70/46 "an expansion of
the 70/45". (McCollister, Tr. 9673-74; DX 672, p. 21.) McCollister
la i testified that the hardware for the 70/46 was "in its elements iden-
17
IS
tical" with that of the 70/45 with "the addition of some faster
registers in the machine". He estimated the hardware development
19 : i effort of the 70/46 was "in the order of $2 million . . . because we
2Q ] made so much use of what was already existing in the 70/45".
21 ; (McCollister, Tr. 9679.) The Spectra 70/61 had a comparable
22 1 relationship to the 70/60 in terms of design approach as
i
<n ! the 70/46 had to the 70/45. (McCollister, Tr. 9680.)
~ i
74, | RCA, like GE and IBM (see above, pp. 417-18, 505-06),
i
2c 1 went into computer time sharing because of the changing demands of the
-567-
r- !:
industry. The introduction of the 70/61 "was in response to accel-
erating industry shift from traditional batch orocessina to r^nte
computing, a system in which a large central computer accepts and
^ jL almost simultaneously feeds back data to numerous remote terminals" .
(DX 674, p. 8.)
As a result of "[t]he growing acceptance of remote com-
puting", RCA foresaw "excellent potential for sales of data communi-
cations terminals and other peripheral equipment as well as for com-
puter hardware" and expanded its manufacturing capabilities for
peripheral computer equipment. ( Id. )
However, orders for the 70/46 and 70/61 during the years
1969-1971 were "less than had been projected in the forecasts because
. . . the 70/46 had originally been planned as a system from which to
gain experience with this class of product". (McCollister, Tr. 9694.)
The marketing forecast was "excessively optimistic". (McCollister,
Tr. 9695.)
RCA, like GE and IBM, ran into substantial difficulties
developing the time-sharing software. RCA's time-sharing software was
r
i
T
3
3
a
2
3
A
.s
LS
•J called the Time Sharing Operating System (TSOS) .* The development of
. TSOS was "[b]y far the largest software development or largest pro-
.} gramming system" that RCA's computer division had undertaken.
21 '\ (McCollister, Tr. 9697.)
22 : ;
^ | * At a later time TSOS was known as VMOS . (McCollister, Tr.
| 97X7-18.)
-568-
1
z
3
4>
5
S
7
3
9
ia
ii
12!
13
IA!
15
16
17
IS
20 '
21
Despite this effort RCA had
"[i]mportant difficulties with . . . TSOS-VMOS which
substantially impaired, certainly in early installa-
tions, the performance of the system as a whole, which
includes both hardware and software." (McCollister,
Tr. 9710-11.)
The problem with TSOS was that "there were bugs in it, which took time
to get rid of, and it was late, as far as providing functions
specified were concerned". (McCollister, Tr. 9694.)* Similarly,
Rooney, who joined RCA in 1969 and was President of the Computer
Systems Division in 1971 and 1972, testified that:
"The Time Sharing Operating System was a form of
virtual memory system that had a great deal of func-
tional capability to offer, that was new and unique ■
in the marketplace, but its reliability in performance
was extremely poor and we had not achieved a high
degree of reliability with that system while I was at
RCA.
"The system was referred to as bombed out'. There
would be a problem. It would essentially go down.
There had been a malfunction in the hardware, but in
essence it was what was referred to as a bug in the
program of the operating system, but it was not able
to cope with handling certain data, as it was speci-
fied to handle it." (Rooney, Tr. 12132-34.)
*
McCollister observed that developing time-sharing software was
ii difficult "for other manufacturers attempting [it]" as well as for
ij RCA. (McCollister, Tr. 9694.) He added:
2- ij "The history of the computer .industry is filled with
^ ,! examples of difficulty that every manufacturer has had
with the introduction and initial installation of new
products, either hardware or software. This is a normal
way of life.
23
24
25
" [S]oftware, in particular, is a complex technical
-569-
L \ As a result
Z
3
4
5
6 ;
7
8
Si
10
11 \
12- J
13 [
i
i
1± !
** i
I
15 ]
"It was available to the user, but there were a
great many periods of down time and, also, if you
were operating with terminals, the response of the
system would be very s«low in a timesharing mode."
(Rooney, Tr. 12134.)
In July 1971 r Rooney identified "inadequate software" as among the
"functional capabilities which detracted from [Spectra's] ability to
meet its product. objectives" . (DX 11101, p. 1.)* While the per-
formance of TSOS/VMOS improved "as time went on" (McCollister , Tr.
9718) , development work on TSOS continued until RCA left the computer
business. (McCollister, Tr. 9674.)
task, and very large programming systems have literally
hundreds of thousands of instructions in them between
which there is interaction, and sometimes you don't
know whether or not there is a fault until for the
first time a particular combination occurs, and it is
very difficult to develop this total mass of logic and
to make certain that it is all correct at the time you
turn it over to a user of equipment. This is a common
^g -J experience within the computer industry.
17 !|
'I
18 i
19 ;«
20
"For another reason, competitive conditions in the
industry have been such that the manufacturers are
always under extreme pressure to get out a new set of
equipment and get it into the marketplace and get it
installed as quickly as they can." (McCollister,
Tr. 9696-97.)
i * We are aware that this document is not in evidence; however,
-^ :', we believe it is reliable and rely on it because it was written
■i by a person with knowledge of the facts (J. W. Rooney, President
22 J of RCA's Computer Systems Division) contemporaneously with the
^ ;j events described and confirms Mr. Rooney ' s trial testimony on
-« i the same subject. (See Tr. 12131-37.)
24 ;i
25 i
i| -570-
i
2
3
4-
5
6
7
3
io i
i
ii i
12
13 i
!
t
i
15 !
i
I
15 I
17 I
13 |
19 i
I
21 ii
Other Product Problems. In addition to problems with its
processing units, RCA experienced problems with the operating software
(in addition to time-sharing software) , random access memory units,
card readers and memory stacks on Spectra 70 . These problems created
substantial problems for RCA in its marketing of Spectra. As Beard
testified, in addition to looking at the price/performance or through-
put claims of RCA, a user to determine the value of a computer system
"would be looking primarily at the total system service
that he would be supplied with, not only the effectiveness
of the hardware to perform his requirements, but what soft-
ware would be made available, and how effective the software
was, the caliber of the maintenance organization to maintain
the equipment once it was installed, his impression as to
how well he would be supported by the vendor on future
applications which he had not yet fully defined in his own
mind in terms of new equipments, new software." (Tr.
10090.)
In addition, the user would take into account the relative functional
capabilities of the peripheral equipment offered by competing systems.
( Id. ) When compared to IBM, the overall "value" of Spectra to the
user did not match IBM's 360.
TDOS. For its Spectra systems from the Spectra 45 up, RCA
I had an operating system called TDOS. Rooney testified:
"When it was announced it was a good system, but
RCA did not continue to improve upon it at the same
pace as IBM improved upon their OS. Our system, while
performing satisfactorily in terms of reliability, did
call for a lot more operator intervention in terms of
22 ';! performing the work than the OS system.
i
22 i "1 made a strong plea for an improved system
j called OS 70, which was under development, to be
ii used on the RCA 6. That system was decommitted in the
z ' ! | early part of 19 71.
25 !
-571-
"By ' decommitted ' I mean it was never brought to
the market.
"The people responsible for putting the system
together felt that they couldn't do it in the time
frame that had been asked for". (Rooney, Tr.
12135-37.)
Peripherals and Memory Stacks. As discussed, RCA intro-
duced its RACE file during 1964, and it suffered from various
problems. (See above, pp. 549-50.) RCA marketed the RACE with its
Spectra series and continued to market it actively into 1968 at
which time it
"was impacted by the progressive development of disk
file technology. Disk files were more reliable
devices. There were fewer things to make mechanical
trouble in them. They had a faster access time and a
faster transfer rate of information from the medium into
the processor, and as the cost performance characteris-
tics of disk files improved, the relative advantage
and cost performance of the so-called RACE unit was
reduced, until you reached the point where, for most
applications, a disk file, as illustrated by the 2314,
was a preferred approach." (McCollister, Tr. 9659-60.)
During the middle 1960s RCA still was not producing disk
drives. To meet the demand from users for disk drives, RCA purchased
; IBM 2311 and then CDC 2311-type disk drives for use with the early
deliveries of the Spectra 45 and 55 in 1967. (Beard, Tr. 9935.) RCA
i did not deliver its own 2311-type disk drive until the end of 19 6 7 or
f !
" :j beginning of 1968, a year and a half after its first Spectra deliv-
1 ! * .
1 eries. (Beard, Tr. 9913.)
When IBM began deliveries of its 2314 disk drive, RCA found
4 | that its marketing people "were under a handicap in selling the Spectra
e i
- :!
I
■I -572-
70 Systems. We did not have a comparable product to the IBM 2314 at
the time." (Beard, Tr. 8575.)
RCA's development of a 2314 equivalent was hampered by the
departure of the group of persons who had worked on development of
its 2311-type disk drive in 1967-68 to form another company, which
g | was called Linnell Electronics.* The departure of those persons
7 j| impaired RCA's ability to develop new disk drives. (Beard, Tr.
g 'it 9924-28.) During. 1968 RCA determined, that Memorex was ahead of RCA
in disk drive technology, and RCA contracted to have Memorex supply
RCA with its "first year or year and a half supply of disks". Obtain-
ing disk drives from Memorex "cost additional money" because RCA
"had in parallel [its] own development going on which was going to
be about a year later than Memorex 1 s". However, RCA could not
afford to wait the additional year because it was "losing too many
sales" to IBM "for the lack of it". (Beard, Tr. 8574-75.)
RCA supplied its own controller for the Memorex 2314-type
drives (Beard, Tr. 10254-55) , developing it at an engineering cost of
: | about $500,000. (Beard, Tr. 10246.) RCA started to work on a 2314-
:| type product as a "full design project" in 1968 (Beard, Tr. 9922) and
13 :;
j delivered the first units to customers "around the latter part of 1969
20 ':
! to perhaps mid-1970". (Beard, Tr. 9915.) The fact that RCA "was not
21 ] ■
:i able to produce on its own or to duplicate the 2311 or the 2314 disk
*2. »
9>
LC
U
12
12
14-
13
16 II
17
■\
" i
j * In 1968 Linnell was manufacturing IBM-compatible 2311-type disk
2- ) drives for use with System/360. (DX 12543.) By 1972 it was manufac-
;! taring 2314-type compatible disk drives as well. Both drives were
25 i marketed by Bryant Computer Products. (DX 4556, p. 2.)
i
■t
"\ -573-
drives until very much after the IBM delivery" hurt RCA. (Rooney, Tr.
12123, 12192-94.) Disk drives offered "a functional capability very
much needed in terms of price/performance in the competitive market-
' place and without that capability you were in a weak competitive
situation against IBM" ► (Rooney , Tr. 12192-94.) Thus, in July 1971
I |j Rooney wrote that "the lack of a 2314 competitive device until late in
the product life" was one of the "many functional capabilities which
detracted from [Spectra's] ability to meet its product, objectives" .
(DX 11101, p. 1.)
RCA also had initial difficulties with the card reader on
its Spectra series ► The difficulties were corrected only at the cost
of making it a "very high cost product". (McCollister, Tr. 9604.)
One of the most severe problems faced by RCA was that
of providing reliable memory stacks in 1967 and 1968. This problem
cost RCA as much as $10 million and caused J. R. Bradburn, Executive
Vice President of the computer division, to write to R. W. Sarnoff,
President and Chief Executive Officer of RCA (PX 338, p. 49) in
December 1968 recommending that the Memory Products Division be
transferred to his division:
I
I
L
1 i
3 I
L
5
5
:{
i
1 "Development processes must involve more than
>A 1 theoretical analysis and its immediate physical
; ! embodiment. A thorough understanding and consideration
'C !
"Modern computing and data processing systems j
£ ; | consist in essence of input/output peripheral equip- j
i ment, control, and memory. Development of complete ;
11 :] competitive systems involve [sic] simultaneous, j
j continuous, and coordinated development of all com- ;
2. ;| ponents. The single most important element of this
j overall development is memory. '
-574-
L
2
3
4.
§
of mechanical design., reliability, manufacturability ,
and maintainability of a complete memory system is
required. Nothing less can meet competition today.
"The present organizational structure within RCA
is not conducive to efficient operation or to meeting
these requirements. It does not bring to bear upon
the decision making, process the needed emphasis or the
proper sense of order of importance adequate to meet
£ |t the needs both in the short and long runs .
3
La
u
12
13
14-
15
IS
" . . * [This] is what has been demonstrated by
the inordinate difficulties 'encountered in trying to
provide reliable memory stacks for our computer ship-
ments in 1967 and 1968 - Poor stacks may have cost us
as much as $10,000,000 in those two years. Additionally,
our problem is portrayed by what has been inadequate
provisioning in the engineering budgets of Memory
Products." (DX 840, p. 1.)
(iv) RCA Success with the Spectra 70 Series. Despite the
numerous problems experienced by the Spectra 70 series, during the
period of its life (1965-1969) RCA enjoyed considerable success with
its computer business. McCollister testified that:
"1966, "67, '68 and r 69 were generally periods of
steady and encouraging growth, and we were operating
for the most part during this period at pretty much a
■| breakeven, although in the year 1966, because we had
W it a heavy installation workload, we upped the budget of
'.I the marketing organization, which threw us back into
-S :j the red, and in the following year or two we began to
increase the amount of money going into the engineering
organization .
19
20 ;j "But I would say that beginning in 196 5 through the
i year 1969 we were making what appeared to be encouraging
21 ; ! - progress. We did have an ability to compete within our
i particular scope of operations and the corporation was
22 1 encouraged about the long term outlook for the Division."
(Tr. 9246-48.)
23 ;|
24>
25
I RCA's annual reports confirm McCollister ' s assessment:
1
■J In 196 4 RCA reported that its "gross computer sales and rentals" were
-575-
higher than $100 million, having grown from $14.6 million in 1960.
"RCA's total data processing business earned a profit for the full
year." (DX 669, p. 2.)
In 1965 RCA reported that its profits in the computer busi-
ness continued and "the potential for future profits was enhanced by
the booking in 1965 of orders for 92 percent more computer systems
than in the preceding year. By 1970, profits from the data processing
business . . . are expected to become a highly significant factor in
RCA's total earnings." (DX 670, p. 7.)
In 1966 RCA reported that "domestic orders for RCA computers
and their associated equipment rose by 53 percent over the 1965 level".
During 1966, with Spectra deliveries beginning, RCA enlarged its field
marketing force by 45 percent, planned for another 35 percent increase
in 1967, and "boosted production capacity by 75 per cent to fulfill
. the growing demand for Spectra 70 computers and other data processing
. It equipment". (DX 671, pp. 4, 15.) RCA reported that a loss in its
'] computer business in 1966 was caused by an increase in leasing as
] opposed to purchasing by customers.* ( Id. , p. 4.)
3 i
:| In its 1967 Annual Report RCA reported that:
ff i
:l * Concerning the change in favor of leasing by customers RCA
- ! reported that:
4
2 I "The increase in lease transactions was common
■j throughout the computer industry and reflected in part
2 j the tightness of the money market. While tending to
promote long-range stability by spreading income over
the period of lease, it reduced immediate income. This
trend, as well as increased spending for future growth,
)A
contributed to a loss in RCA's computer business ror
i 1966."
•t
} -576-
Li
2 j
I
i
si
»
i
7 I
t
si
i
i
*!
10
i
111
i
i
12 |
13!
14
15
16
17
13
19
20
21
22
23
24
25
"Our computer business continues to grow at a faster rate than
that of the industry as a whole. However, because most of our
equipment is leased, rather than sold outright, income is neces-
sarily delayed. In 1967, this situation placed us once more in
a loss position, but we look upon this as an investment in future
profits and we look forward with confidence to the period when
our data processing activity will become one of the most importair
parts of our business, surpassing even color television."
(PX 338, p. 5.)
In 1969 RCA reported:
"Domestic bookings of RCA computers were more than 40 per cent
greater than in 1968 and represented nearly a threefold growth
in the past five years.
"As the decade ended, our backlog of computer orders was
30 percent higher than a year ago. The trend toward remote
computing systems as well as the mushrooming of data processing
applications make the outlook for the '70s very promising. We
expect the dollar value of our information system shipments to
increase significantly during the 1970' s and to approach $1 bil-
lion annually.
"While total revenue from sale and lease of computers
rose 23 per cent during 1969, RCA's computer operations remained
in a loss position. This deficit is largely the result of
expenditures aimed at future growth, which include the building
up of our marketing forces and expansion of software and other
aspects of the business." (DX 674, pp. 8-9.)
During the period 1965-1969, RCA's U.S. EDP revenue rose
from about $89 million to approximately $211 million. (DX 8224, p. 2.
RCA added
"A pattern of fluctuating profits and occasional losses is to be
expected in the development of a strong base for the future in
this complex and competitive field." (DX 671, p. 4.)
-577-
c# RCA Computer Systems Division 1969-1971. The story
of RCA's involvement in computer s at the end of the 1960s into the
early 1970s has three parts:
First t the change in personnel at the corporate level and
in the computer division during that time frame and resulting
changes in goals of the computer business;
Second , the decision to develop and the consequences of
the RCA Series ; and
Third, the problems that resulted from the preceding two
decisions.
( i ) Changes in Management Personnel and Goals . On
January 1, 1968, Robert Sarnoff became Chief Executive Officer of
RCA while continuing as President; his father, David Sarnoff,
continued as Chairman of the Board. (PX 338, p. 49.) During 1963
Chase Morsey, Jr. left Ford and joined RCA as Vice President of
Marketing, and the next year became an Executive Vice President of
RCA. (McCollister, Tr. 11156-57; PX 338, p. 49; PX 339, p. 49.) In
1970 Robert Sarnoff replaced his father as Chairman of the Board
\ (DX 674, p. 36), and in 1971 A. L. Conrad became President and
'; Chief Operating Officer. (PX 341, p. 38.) Conrad had worked his
T i
way up through the RCA Service Company (PX 400) which, " [i]n addition
: 4 to its work for the government and in education, the Service Company
| install [ed] and maintain [ed] home-entertainment products, commercial
3 ;
-578-
I.: electronic systems, and business and industrial equipment". (DX
2 : 677, p. 13.)*
3 | At the time Robert Sarnoff became RCA's Chief Executive
4»|[ Officer, RCA underwent a change in its corporate philosophy. In its
1968 Annual Report RCA reported that:
3
a
7
3
9
10
II
12
13
15
"In its formative years, RCA's growth depended primarily
on a single product or service.
"The word that best characterizes the modern RCA is
diversity. An almost even balance has been achieved between
manufacturing and service operations, and the well-being of
your company no longer hinges upon any single activity.
Carefully planned diversification has moved us into new
areas of enterprise, such as vehicle rentals through The
Hertz Corporation and book publishing through Random House.
New businesses have been created from within, such as the
Information Systems Division". (PX 339, p. 11.)
RCA's desire for diversity caused it to acquire many
different and unrelated businesses. Starting in 1966, RCA acquired
the Hertz Corporation (automobile rentals) ; Random House and Ballan-
^ :! tine Books (book publishing) ; Coronet Industries (carpets) ; Banquet
17 ; Foods (frozen foods) ; Oriel Foods and Morris & David Jones (wholesale
IS ; food distributors in the United Kingdom) ; Cushman & Wakefield (real
13 1
20 i * RCA's changes in higher management have continued to the
j present. On December 31, 1975 Sarnoff resigned as Chairman of the
21 ! Board of RCA. (DX 951, p. 36.) He was replaced as Chairman during
j 1976 by Conrad who remained as President. (DX 13852, p. 37.) In
22 :i September 1976, Conrad resigned as both President and Chairman, and
Edgar Griffiths replaced him as President, while the position of
Chairman remained vacant. ( Id. ) On January 1, 1980 Griffiths
! filled the vacant position of Chairman and Maurice Valente became
24. j President. (DX 13902. p. 48.) In June 1930 Valente was forced to
i resign as President, and RCA abolished the position and created an
25 j Office of the Chairman consisting of Mr. Griffiths and five other
f j current executives. (DX 13861.)
(
;| -579-
i
i
.i
estate) ; Alaska Communications System (communications) ; and a color
tube manufacturer in the United Kingdom. (Conrad, Tr. 14002-05;
DX 671, p. 5; DX 674, p. 21; DX 677, pp. 14-19.)*
During that period it entered several new businesses
including the domestic common carrier satellite business. It also
undertook, significant new investments in Colortrak (an advanced
color television receiver) , SelectaVision, VideoDisc, Global Communi-
cations and the RCA Service Company (related to lease and maintenance
services for private telephone systems). (Conrad, Tr. 14002-05.)
Thus, by 1971 RCA was a conglomerate engaged in a large
variety of different fields including home appliances, televisions,
radios, recording devices, federal defense contracts, communications
services, broadcasting, automobile rentals, food, carpets, books,
records, and real estate — and computers. (Rooney, Tr. 12022-24.)
During the 1966-76 period, the chief executive officers
!j of various RCA divisions and subsidiaries, including Banquet Foods,
Coronet, Hertz, Random House and Global Communications, were directors
i i
i
l ;
i
■L
7 ;
s]
.1
of the RCA Corporation. No officer of the computer division was a
member of the RCA Board of Directors during this period. (Conrad,
| Tr. 14027-28; PX 339, p. 49.) In 1971 revenues from the operations
■Q
| of the RCA Computer Division represented less than 10% of RCA's
II : |
?2. ■•;
22 ■ * Bv 1979 RCA had sold Alaska Communications System and Cushman
] and Wakefield. (DX 951, p. 17; DX 13902, p. 14.) As of 1980, RCA
Z*. j had sold Random House and was trying to sell Banquet Foods. (DX
■J 13860.)
-580-
J- i corporate revenues — $270 million out of $3 billion. (Rooney, Tr.
12025.)
At this time changes were also occurring in the personnel
and goals of RCA' s computer division. During 1969 and 1970 RCA
hired people from IBM to manage parts of the computer division. For
example, L. Edwin Donegan, Jr., became Vice President of Sales in
1969 and General Manager in 1970. (McCollister, Tr.. 11590.)
Joseph W. Rooney came from IBM in 1969 and after a brief corporate
staff job became Vice President of Marketing; in 1971 he was
President of the Computer Systems Division. (Rooney, Tr. 11687-88.)
V. Orville Wright was hired in 1970 and the next year was head of
2
3
4-
5
S
7
8
9
La
u
£2 i! Systems Development. (Wright, Tr. 12785.) Sam Adams was responsible
13
14- !
16
for business planning. Bill Acker was put in charge of the financial
operation. (McCollister, Tr. 11590.)
The goals of the computer division changed with the new
corporate and computer division management during this time. Until
.- II Robert Sarnoff took over in 1968, the RCA computer division had
TS j placed its emphasis on accomplishing its business plans and obtaining
ii
ta 1 moderate growth, an emphasis which McCollister considered "correct".
_ \ (Tr. 11156-57.)*
20 i
,1
22j * Withington similarly identified and endorsed RCA's emphasis
1 during that time :
23 I
"In 1965, the RCA strategy was to offer a line of general
24 | purpose computer systems . . . with instruction set com-
j patibility to IBM processors, . . . and to grow at a
25 :; modest rate commensurate with maintaining profitability
at all times.
-581-
In 1968 RCA changed its emphasis to one of quickly obtaining a
larger market share. McCollister attributed this change to the
newly-arrived Chase Horsey,* and testified that the change was not
"beneficial" to the computer division because:
"It tended to place the emphasis upon increasing market
share and relatively deemphasize control of expenses and
achieving a profit, and the end result is that the expenses
in the RCA Computer Division mounted to the point where
they contributed significantly to RCA's withdrawal from the
business.
"In other words, you place the emphasis upon share of
market and you tend to deemphasize some of the other important
aspects of running a successful business, and share of market
is only one consideration". (Tr. 11158.)
In its 1970 Annual Report RCA reported that:
"Our highest priorities today are the establishment of
a profitable computer business and the capture of the
domestic industry's No. 2 position. RCA has made a greater
investment in this effort than in any prior venture in its
history, and we are convinced that the returns will be sub-
stantial". (PX 340, p. 3.)
This change in strategy was visible outside RCA. Withington
> ;! testified that:
r •{
'I , —
"That was RCA's historic strategy since the announcement of
the Spectra 70 series, . . . the system seemed versatile and
attractive, and I saw no reason why RCA could not continue
growing at a modest rate and increase its total market and
perhaps even its market share on the basis of its current
offering". (Tr. 56702-03.)
* McCollister believed that, having come from the Ford Motor
Company, Morsey "was very conscious of share of market statistics
and he was also influential, and this caused the division to give
increasing recognition to share of market as such". McCollister
regarded Morsey 's emphasis on share figures as "exaggerated" and
"a legacy from his experience in the automobile industry". (Tr.
11156-58.)
-582-
"... RCA's new strategy involved attempting to grow very
sharply in market share from the position they were in by
means of attempting to capture IBM 1 s customers in the processes
of evolving from the 360 series to the 370 series . . . RCA's
new strategy involved expanding its operation in every respect, [*
field organization, the manufacturing capability, and the
engineering force, at a substantial increase in cost; and more
importantly, it involved the anticipation of incurring the
manufacturing costs of a very large number for RCA of new
systems in a short time. . . . These financial requirements
would inevitably cause a lack of profitability for a period".
(Tr. 56707-09.)
It was this change in strategy that led directly to the
ill-fated RCA Series.
(ii) The RCA Series . During the 1968 time frame RCA
realized that if it was to achieve its new growth objectives it
would need successor systems to the Spectra series. Unfortunately,
RCA was undecided about what successor products to develop.
McCollister testified:
"In looking at the next family or generation of equipment
beyond Spectra, there was a lengthy debate between the people
responsible for programming systems, that is, the so-called
software organization, and the people responsible for hardware
or equipment specifications , and perhaps the engineering
organization as well, as to exactly what the nature of this
product should be." (Tr. 9809.)
19 !
'» line. The first attempt, referred to at trial as the X series, was
decommitted in 1969 for two reasons: first, RCA felt that it could
20 ;.
21
22 !
23
24
25
* For example, during 1971 the computer division expanded its
operations: it opened a sales office in the United Kingdom to
market RCA computer products in Europe (Rooney, Tr. 12365) and it
constructed a new manufacturing site in Marlboro, Massachusetts.
(McCollister, Tr. 10963-65.)
-583-
not meet what it predicted to be IBM's announcement of 370*; second,
the series included
" Cain architectural problem [in] that they doubted they would
ever be able to complete the product line without a major
restructuring of their whole development program. ... My
understanding at the time was that they could not build it at
ail if they had developed it or had set up the architecture".
(Rooney, Tr. 122^5-26.)
The second attempt at a successor to the Spectra was the
New Technology System (NTS) . The NTS was originally scheduled for
announcement in early 1971. However, " [tjhere was a slippage in
that program and it was subsequently put off for announcement for
approximately 18 months as a result of development problems within
RCA itself". (Wright, Tr. 13173.)
RCA made only "marginally small" investments in NTS.
(Withington, Tr. 57079.)** It appeared that NTS, if announced as it
was being developed, would have encountered competitive difficulties.
Withington testified:
"The basic reason for my concern was that I believed at
the time that IBM would introduce a new family of general
j !( purpose computer systems in the time frame 1973 to 1977, which
It was the time frame in which RCA's NTS computer systems were to
- J be shipped.
9
* This prediction was based on "intelligence" from people who
£ I had worked at IBM and from trade journals. Based on this intelligence
i RCA understood that IBM announcements would begin in the second or
n | third quarter of 1970, with larger models announced first, and
- : ; announcements continuing throughout 1970 until the lower models of
•! the line would be announced at the beginning of the 1971 period.
— } Shipments were anticipated to be 12 months later. (Rooney, Tr.
12225-29.)
.2
15
** During this time RCA hired A.D. Little to review RCA's product
and marketing strategies in its computer business. (Rooney, Tr.
11814.) On behalf of A.D. Little, Withington made suggestions and
wrote a report. (DX 2666.)
-584-
^ t
i
z\
4*
5
6
7 1
3
9-
Id
II
12
13
14
15
IS i
17 |
IS I
19 i
>
ZQ j
21 I
"I believed that the nature and functionality of the NTS
line would be inadequate to meet the needs of customers who were
IBM users or who would otherwise consider IBM systems during that
time frame."*- (Tr. 56715-16.)
Withington told RCA that "a revision and acceleration of
the product plan would be necessary if RCA would have . ► . ' [a]
good chance of attaining the desired market share'".** (Tr.. 56716;
* Withington listed the ways in which the NTS line would
be "inadequate in. terms of functionality":
"I noted that RCA's NTS line involved only mono-
processors rather than multiprocessors. I noted there
would be no small satellite processors, and I noted
that NTS was to be aimed at the medium monoprocessor market
segment, this meaning in terms of price range, and noted my
belief that the IBM customers would be migrating from the
medium monoprocessor market segment to large multiprocessors
to obtain the advantages of centralization.
"I foresaw that the IBM systems of the 1973 to 1977 range
would be equipped with multiprocessing, satellite systems,
virtual memory, and fail-soft, of which all but the latter
were in fact announced; and that there would be a trend toward
centralization, which did take place.
"Since RCA's NTS line did not fit with either that
functionality or that trend, I believe that was the entire
reason for my concern". (Tr. 56715-16.)
** He recommended that RCA embark on a development program for a
satellite processor and on a "comprehensive terminal plan" because
"an adequate line of inquiry interactive and remote batch terminals
must be offered with network systems": because of their "visibility"
and large numbers, "terminals may dominate users' selections". The
H existence of a terminal product line for RCA "might very well" have
22 | made a difference in the success of the NTS plan because "the
j profitability of the computer business as a whole might have been
23 ij greatly enhanced" if RCA had had excellent terminals. (Tr. 56718-
20; DX 2666, pp. 5, 15-16.)
24 i
25
i
-585-
L' see also DX 2666, p. 5.)
? During 1969 and 1970, RCA. looked at other ways to replace
its Spectra Series:
"One was to specialize in just one or a few industries
and to market in no other than those few industries .
Another alternative was to select an individual niche in
the product line by rental size where we could market
exclusively products in that area". (Rooney, Tr. 11820-21.)
7 | At this point, during 1970, RCA had several alternative
•:i
a :
3
10
il
12
-a
14-
16
17
ways it would proceed in the computer business : it could continue
to market the Spectra 70 series until NTS , or some other more
advanced product line was developed; it could specialize in a
particular product area or it could market what became the RCA
Series. The latter option — the RCA Series — was chosen for several
reasons :
(1) The new management of the RCA Computer Systems
Division wanted to stop marketing the Spectra and to market
its own line of products:
"The then management of the Division wanted to
have a product line that would be associated with their
: j management era or period, as opposed to a product line
IS ;j which was associated with an earlier management era".
(McCollister, Tr. 9837-38.)
19
.:
i
il i
•i
12 !
t
It also thought that a new product line would have a
"psychological influence" on the "marketplace". (McCollister,
Tr. 9816-17.)
(2) RCA also believed that it could not continue to sell
the Spectra series in the face of the price/performance
improvements offered by I3M with its System/370. According
-586-
I
2
3:
4
5
S
7 i
3
9-
10
11
12!
13 j
L4- j
i
15 ;
i
16 j
17 I
IS j
19 j
2Q ;
21 j
22 I
24 :
25
to Rooney, if RCA had not been "selling against IBM" it could
have continued to offer the Spectra series. (Tr. 12234-36.)
(3) Because of its desire for a large market share RCA
rejected the idea of focusing on particular product areas:
" [A]t a meeting I attended, [A.D. Little] presented the
concept that you had to have a broad product line because
you could not possibly sell enough share of any particular
product category to achieve this goal and that strategy
was accepted as being valid" by "[t]he management of the
Computer Systems Group as well as corporate management" .
(Rooney, Tr. 11814.)
(4) RCA believed that it could equal or better the
price/performance of the IBM 370 systems and take away IBM
users by introducing the RCA Series :
"We were faced with a pending IBM announcement; we knew
that the IBM announcement would offer their clients
improved price/performance; we had just had the X series
decommitted; and our objective was to grow to 10 percent
share of the market. And we felt that we had to therefore
maintain our original strategy of going after the IBM
base. And after many discussions, it was concluded that..
by putting in the new memory capability we would be able
to bring the cost of these systems down, so that we could
offer a price competitive system — price/performance
competitive system with IBM's 370. And since it was
following the Spectra architecture, conceptually it would
be the same strategy as IBM was employing, that is,
utilizing the existing software for the next generation
of equipment". (Rooney, Tr. 12242-43.)
The RCA Series was announced on September 15, 1970, three
months after the first announcement of IBM's System/370. (Wright,
Tr. 13175-76.) The RCA Series consisted of four models "of small -
to-medium-class computers — RCA 2, 3, 6 and 7". (PX 34 0, p. 17.)
RCA described the RCA Series as "offering more power and
memorv for the dollar than present third-generation svstems". (DX 675
-587-
I
p. 16.) While the RCA Series had new memories, "under the covers,
the RCA Series was essentially the Spectra 70". (McCollister, Tr.
9819-20.) McCollister described the RCA Series:
M [lit was a restyled prqduct line. There was a new set of
covers,, the frames were the same, and it was essentially a
cosmetic treatment of the existing Spectra 70 Series with
new model numbers and new pricing.
"There may have been some minor improvements. But
'J fundamentally the product was not changed from the Spectra
\ 70." <Tr. 9816-17, see Tr. 9819-20.)
\
}
1
L S
I I
2
&
5
3
,7 .,
'[paperwork and many presentations with respect to this rationale".
:{ (McCollister, Tr. 9838.) Similarly, Rooney testified that in 1969-
a J
\ 1971 "we had a term called intercept strategy, which implied inter-
*f* !
'Jj •
! cepting the upward migration of the IBM client base with RCA equip-
II S .
jment". (Tr. 11811-12.)
22 ; !
i The "elaborate strategy" failed.* The RCA Series was a
RCA's computer division management devised "an elaborate
strategy" to make the RCA Series succeed:
"There was a very elaborate strategy at the time as to
where these units of the RCA series would fall against the
IBM either 360 or 370, either as it had been announced or
was expected to be announced, and I think there was a
fallacious expectation that in this elaborate strategy
that the RCA series would fall at a certain point within
the IBM product line spectrum and that IBM would be
unwilling to disturb the equilibrium of that product
spectrum and, therefore, negate the rationale of the RCA
product concept." (McCollister, Tr. 9837-38.)
Under this "elaborate strategy" the RCA Series would "intercept" the
System/370. "Certainly there was a great deal of effort and much
24 1
* At the time of the decision Withington cautioned RCA about the
25 | RCA Series. He testified:
.1
1 -588-
"major product failure" and "a mistake" (McCollister, Tr. 9819-20;
Wright, Tr. 13577-78; Withington, Tr. 56454-55.) The RCA Series
failed in two respects: (1) instead of "intercepting" System/370,
it "intercepted." RCA's own Spectra 70 series, and (2) it had sub-
stantial technological problems.
Interception of the Spectra 70. The witnesses used
different words but all said the same thing: by introducing the RCA
Series, RCA. "obsoleted" , "intercepted" and "blew . . . out of the
water" its Spectra 70 series. (McCollister, Tr. 9838-39; Withington,
Tr. 56720.) Withington explained what happened:
"[A] great number of present Spectra 70 users renting . . . their
systems, did indeed order the new RCA Series and indicate
their intention to return their Spectra 70s, and that
phenomenon indicated immediately that the financial impact
of the new series would be more negative than planned"*.
(Withington, Tr. 56711; see also PX 4836, p. 23.)
So did McCollister:-
" [A] customer had everything to gain by ordering an RCA
series and returning the Spectra 70 . He got a brand new
L
2
3
*
7
a
11
12.1
12
14
15
is :! '
v J
\\ "I remember indicating my concern to RCA management at
XS \ that time about the likelihood that the new RCA Series would
;j cause a general replacement of rented Spectra 70 machines in
£9 : ;| the field, and that this would cause more negative financial
;[ results than they were expecting. And I remember being
2Q j uncertain about the degree to which IBM customers would be
;! willing to convert quickly and in large numbers to the RCA
2X \ Series". (Withington, Tr. 56710.)
22
22
Withington "did not reach a firm conclusion at that time that
failure was inevitable". He testified that the strategy might have
been successful "if it had been carried out differently with different
prices for the machines, with less effort to grew suddenly, and with
2-4 ; j more effort on functional improvements in the product line". (Tr.
56710-11.)
25
\ -589-
machine. It cost him maybe 15 percent less or so and why
not?" (McCollister, Tr. 9837-39.)
The interception of the Spectra by the RCA Series seriously
hurt. RCA in several respects. First, it reduced RCA's rental
income because rents for the RCA Series were lower than for the
Spectra. McCollister testified:
" [T]he announcement of the so-called RCA Series of
equipments as replacement to Spectra 70, which offered new
equipment under different model numbers, which was technically
essentially unchanged from the previous equipments , but which
was offered at a lower price than previous equipments, meaning
those equipments which were in that rental base . . . hastened
the return of those equipments from the rental base to the
manufacturer.
"The effect was that it hastened the return of equipment
that was then installed with customers on a rental basis
because the customer could get a newly manufactured machine of
equal ability, a new appearance and at a lesser price than he
had been paying for the one which he already had installed" .
(Tr. 11491-94.)
Second, RCA was forced to build more RCA Series machines
while it built up an inventory of returned RCA Spectra 70 's.
Wright described the situation confronting RCA:
"You therefore were confronted with the building of new
products by the manufacturing organization and shipping those
products to the installed customer base, and in many, many
instances , because there was improved price/performance in the
RCA Series over the Spectra series, you replace your own
equipment and you got the Spectra series of systems equipment
back. In many instances the cost of manufacturing had not
been fully amortized, and that would have an effect upon both
your cash requirements and also upon your P&L". (Tr. 13577-78.) *
25
McGollister testified that RCA found itself with:
2.J "lo]rders for the RCA Series which required a manufacturing
investment, in the product being placed out on rental for the
3;| most part, which drew capital from the corporation to do this,
and it resulted in the displacement of existing Spectra 70
4. |f processors in many cases before they had been fully depreciated"
(Tr. 9818.)
5
"This tended to build up an inventory of the equipment
7 j| which was returned by the customers, the rental income from
that equipment ceased and the company was faced with the
g j requirement to invest money in new equipment to place in the
customer's office to take the place of that which was sent
g. I back or returned". (Tr. 11491-94.)
IC
He added that:
In addition, the early returns of the Spectra 70s:
*, "[h]ad serious adverse financial effect upon the Divison
because it did not permit us to follow a plan or have a strategy
•2 j which would maximize the return from the investment in Spectra
j 70 equipments." (Tr. 9838-39.)
13- '
14-
15
IS
17
13
19
20;
21 l|
I
22 ;|
55 i
" , above) , by cessation of rentals from the returned machines and the
• necessity to spend money to manufacture the new ones.
25
-591-
' I
The problems of returned Spectra 70 systems were such that
RCA established a Returns Task Force, which made its presentation in
early August 1971 and, considering both returns experienced to date
and those which were forecasted, concluded that "approximately 70
percent of the returns were being caused by RCA's replacement of
Spectra series with RCA equipment" . Approximately 18 percent were
losses to competition and about 12 percent due to economic problems .
(Rooney, Tr. 12275-77; Wright, Tr. 13581-83; DX 873, p. 24.) These
early returns particularly hurt RCA's profit and loss statements by
forcing RCA to write off the undepreciated asset value of its "accrued
equity contracts". (McCollister, Tr. 9820-21.) These contracts were
arrangements in which the customer leased the equipment for five
years, making equal monthly payments over that period, but RCA took
70 percent of the revenue that it expected to achieve into its
profit and loss statement in the first year of the contract. (Wright,
Tr. 13589-96.) When the RCA series was announced, some equipment
under accrued equity contracts was returned prematurely (according
a ;| to McCollister) because the manager of the division "was anxious to
make a showing with respect to the success of this new product line"
and had "an inclination to allow customers to return Spectra 70
t
, j equipments prematurely for the sake of being able to cite an order
^ i! for a machine in the new product line". (McCollister, Tr. 9820-21.)
.- :« This meant that debits against current revenue had to be recognized
J i
, 4 \ when machines were returned before earning the revenue already
»•» i
I
re l
-592-
^ i reported in prior years.
2 ; " Technological Problems of the RCA Series. The RCA Series
3 ; suffered front technological deficiencies that hampered its success:
i •
4-1' "Because you were: in a sense perpetuating technology
J that was five years, old, you were making a new investment
s ! in five-year old technology, and the pace of technology
I in the industry, in it [sic] cost effectiveness charac-
8 I teristics, is such that when you bring out a product line
you cannot afford not to take advantage of improvements
7 1 in cost performance and capabilities up to the time that
you bring out that equipment.
a !■
"That is certainly one aspect of the problem, that it
9 | resulted .in new investment in old technology when better
I technologies were available at that time". (McCollister,
IQ | Tr. 9819-20.)
I
j^ i In addition to the general technological staleness of the
!
M J RCA Series, particular products were deficient in various ways.
I
£2 , Peripherals continued to be a problem. By 1970 RCA was "[t]wo to
T<L ! three years" behind IBM in the development of peripherals.* (Rooney,
• as L Tr. 12247-48.) More specifically, RCA was hindered by its failure
t „ !j to have a disk drive competitive to the IBM 3330, which had been
\ announced in June 1970 with System/370. In 1971 Rooney reported
_ it that RCA still suffered from its:
;| "[i]nability to provide a 3330 competitive device until some
*£ ;| 19 months after IBM's delivery of its 3330 unit. I feel both
, of these items are of major importance to the success of our
20 j rca series marketing efforts and should be resolved. In par-
I ticular, I am concerned we may suffer the same exposure we
21 ! have faced with the 70/564 and 70/590 disc programs if we
4 are not able to accelerate the present delivery schedule
22;i for our 8580 unit." (DX 11101, p. .2.)
2i j
I * Rooney added that there had been a pattern in the RCA experience
25 j up to 1970 of producing essentially carbon copies of IBM's peripheral
! equipment two or three or more years late, either by developing that
1 equipment themselves or by acquiring that equipment from other equip-
,| ment manufacturers. It was one of his goals in 1970 to try to do some-
! thing to overcome that disadvantage. (Rooney, Tr. 12249^50, 12252-53.)
:: -593-
► }•■
I i
f i
And in July 1971, RCA's computer division monthly report stated that
"sales of the RCA 6 and 7 have been and will continue to be hampered
by the large delivery differential between the RCA 8580 [disk drive]
(March 1973) and the IBM 3330 (August 1971)". (DX 11099, p. 5.) By
then RCA was arranging to purchase the 3330 on' an OEM basis from IBM.
(See DX 937.)
In mid-1970 RCA established a Peripheral Task Force to try
to do something about its problem with peripherals. (Rooney, Tr.
12249-53.) On July 16, 1970, L. E. Donegan, Jr., Division Vice
President and General Manager of the Computer Division, wrote to
W. w. Acker in Finance and Administration concerning the activities
? I of the Peripheral Task Force:
I 1
8
"Conceptually, the thing we must begin doing and you
should attack it across the board in all peripheral areas,
is get away from the pattern of producing carbon copies of
IBM's peripherals, one generation late. Now that we have
our hands on some very good IBM intelligence we should
attempt to leap-frog and get ourselves to within 12 months
of their delivered peripheral capability.
. "... the tape units that we are presently developing
' : in Marlboro are competitive in price performance and specifi-
cations with Third Generation IBM tape stations and most
® li likely will be non-competitive once IBM makes its new tape
'( familv announcements. We must keep this from happening."
9 1 (DX 862.)
H
RCA's problems with software also continued into the RCA
- :; Series . VMOS 4 , an operating system to be used with the RCA 3 and
;■!
— j RCA 7, was announced in September 1970. (Rooney, Tr. 12335.) By
.- :»
a j December 1970 it appeared that there would be a 6-9 month slippage
>
-■* 1 in the delivery of VMOS 4. According to A. L. Fazio, RCA's Manager
i
£5 I of Virtual Memory Systems, such slippage would be a "product disaster"
-594-
1
z
5
6
causing RCA to lose about $3.5 million from delayed installations
and approximately $2.1 million points (dollars in monthly rental)
from current and future prospects. (DX 872 , pp. A, B, C.)
The slippage occurred and was "significant". (Rooney,
Tr. 12349-50.) Because of that slippage RCA lost $3.5 million in
revenue and marketed about 40 systems — 20 RCA 3s and 20 RCA 7s — less
j\ than it would have during 1571 through 1973.*
(iii) Computer Systems Division's Problems — Early 1970s.
1970 was "essentially" a breakeven year for the RCA Computer
Systems Division. At the time RCA projected that 1971 would
also be "breakeven", with 1972 showing a $25 million and 1973
a $50 million "pre-tax operating profit". (McCollister, Tr. 9814-15.)
9-!
10 |
U. \
i
tt|
■J3 || A five-year business plan drawn up by the Computer Systems Division
in late 1970 provided for "a breakeven position in 1971". (PX 20 8,
p. 1.) In its 1970 Annual Report RCA painted a similar picture for
its shareholders :
14
IS
16
17
IS IS
20 j
:i
2 1 •!
■
22 ;
23 !
24 i
2s ;
"This investment has already resulted in a more rapid
growth rate for RCA than for the domestic industry as a
whole. In 1970, the value of RCA's net domestic shipments
rose by more than 50 percent, while that of the industry
fell by more than 20 percent. Among the factors responsi-
ble for RCA's progress was a decision to continue increasing
the computer marketing force during a period when many others
in the industry were retrenching as a result of the weakness
of the economy.
* The purchase revenue on each RCA 3* sale lost would have been
$1 million and on each RCA 7 sale lost would have been $2 million.
(If rented, monthly revenue would have been 1/5 0th of the sales
price.) (Rooney, Tr. 12352-53.)
•' -595-
I
2
3'
4.
•i
5
S:
7
3
9
L<3
U
12
L3
\A
Is
16 ;l
17 j
"Our computer revenue this year will be more than
double that of five years ago, and we continue on target
toward a profit crossover in computers in the early 1970' s".
(PX 340, p. 3.)
However, during the beginning and middle of 1971 it was
«
becoming apparent that those plans for the Computer Systems Division
would not be met, and that the Division would be far less successful:
(1) In late January 1971, Robert Sarnoff was informed by
RCA's auditors, Arthur Young & Co., of a "major . . . change
in operating results of the Computer Systems Division in the
1971 business plan", and he asked for an analysis of the
problem. Arthur Young responded by a letter dated February
24, 1971. (DX 11108, p. 1.)
(2) In April 1971 RCA revised its business plan for
the Computer Systems Division. That plan reduced the pre-
diction of revenue for the Division set forth in the December
1970 plan from $323 million to $261 million. The revised plan
predicted an. anticipated pretax loss in 1971 of $37 million.
(DX 952, pp. 7-8, 12.)
IS
(3) On April 23, 1971, H. L. Letts, RCA's Senior
19 ]
:\ Financial Officer wrote to Sarnoff that the magnitude of
20 : !
j the Division's problems raised serious long-term concern
21 1
j about the business and suggested reappraisal of the Division's
22 ]
objectives. He suggested that a task force be set up to study
2a -i
! the Division and its objectives. (DX 952, pp. 1, 2.)
24 \
j (4) By June 1971 a task force comprising six persons
25 -i
1 from Arthur Young and the RCA Auditing Staff had reported
: i -596-
I
2
3 !
5
e
7
3
9-
10
LI
12 |
13 |
14 !
15 !
j
17 j
IS
on the problems of the Computer Systems Division. Morsey
sent this report to Sarnoff and Conrad with a cover memo-
randum stating that "the Computer Systems 1971 loss could
deteriorate significantly from Business Plan levels" ..
(DX 955,- p. A.}
(5) During 1971:
"[lit became apparent that there would be a loss
in magnitude of $30 million or $35 million, a loss that
eventually rose to the area of $50 million or $60 million,
and this of course was an enormous difference from what
had been anticipated at the beginning of that year . "
(McCollister, Tr. 9814-15.)
(6) By the middle of 1971 problems at RCA "put in ques-
tion the anticipated revenue, and in turn opened the question
in my mind as to » . , profitabilitv ... in the remainder nf
1971". (Conrad, Tr. 14125.)
Those and other participants in or observers of RCA's
computer business pointed to many problems in RCA's Computer Systems
Division that would cause its anticipated losses. A discussion of
each follows:
!
i Declining Revenues. The higher-than-anticipated returns
19 1
: from Spectra equipment ($155 million as opposed to $90 million)
• :! resulted in a reduction of expected net shipments (even though
*•* 'I there was an increase in gross shipments) from $230 million to
*- ij $186 million. RCA's 1971 Business Plan stated that:
" Ij " [t]he returns implications of RCA's first introduction of
ii ... the RCA series compatible with the Spectra series . . .
2~ j were not fully reflected in the first plan. . . . There has
; | also been greater migration than expected from the old to the
~ w ;| new series. . . . The increased dependence on the RCA series
■I -597-
:l
H
ij
Ii
has a profound impact on revenue projections and attendant
risk, since product will not be available until second half
of 1971". (DX 952, pp. 7-8.)
Expenses Too High. RCA's revised plan in April 1971 sug-
gested possibilities of actions "to minimize corporate investment
until development of larger revenue base is obtained", including
the deferral of the Marlboro office building, "improved asset
utilization and management", "further expense reductions" and a
possible merger with or sale to another company. (DX 952, p. 21)
McCollister attributed RCA's losses to the fact that
"there was a substantial increase in expense, and that revenues
were not increasing, and that revenues had been seriously over fore-
cast. " An example was the construction of the Marlboro facility
in 1971. "The relocation of the offices, the executive office and
|j the construction of the office building under the circumstances was
... a mistake. ... [It] was an expenditure which could have
been deferred". (McCollister, Tr. 10964-65, see also 9814-16.)*
RCA also "had a very serious problem relative to manufac-
turing costs". According to Wright:
.1
24 1 * The expenditure for Marlboro amounted to about $25 million for
•I capital facilities, not including relocation expenses. (McCollister,
25 j Tr. 10966-67.) McCollister also testified that the 1971 opening of
j RCA's sales office in the United Kingdom was "a mistake" because "it
•\ was an investment which would, if ever, be financially rewarding only
'I at some point in the future". (Tr. 10 96 5.)
;l
J
i! -598-
I
2
3
9-
ia
II
12
13
1*
15
16
17
IS
19
20
21
24
25
"There was no value engineering work going on after
the product was developed to reduce its cost within the
manufacturing organization. Those types of things .
"The cost, as I recall it, when I first got involved
in that, which would have been early in 1971 . . . was
running* at that point in time about 42 percent of revenue."
(Wright, Tr. 13559-60.)*
Poor Organization and Unreliable Information. The
Arthur Young report to Robert Sarnoff in February 1971 reported:
"The basic failure to develop acceptable planning
information in the division involved the lack of a reliable
information base, principally relating to revenues, from
which plans could be developed and current performance
measured. This situation was aggravated by communications
gaps which developed in a period of organizational change.
Planning responsibilities and assignments were not clearly
defined. As an example, the financial group was divided
early in 1970; moves to upgrade the remaining group were
less than successful. Preparing for the move to Marlboro
was probably a further complication." (DX 11108, p. 1.)
The Returns Task Force in 1971 reported that the Computer
Systems Division suffered from the inadequate tracking of computer
equipment :
"1. No single, reputable data base for customer/equip-
ment information.
"2. No two data sources agree.
*
Wright recalled the comparable figure for IBM's manufacturing
cost as a percent of revenue as "on the order of 14 to 15 percent" and
"in certain other products, such as the CPU alone . . . substantially
under that". RCA looked at other companies besides IBM and concluded
ii that Sperry Rand's manufacturing cost "was running about 24 percent
\ of revenue," and Burroughs "was about 21 percent of revenue."
(Tr. 13560-61) When Wright got to RCA he "took several steps to
reduce those manufacturing costs." (Wright, Tr. 13563) For example,
RCA substituted "high quality plastic" covers for the "very heavy
steel gauge covers" it was using at 15% of the cost. (Wright,
Tr. 13564.) This and other cost cutting measures reduced manufac-
turing costs by "8 percentage points as it would relate to revenue".
(Wright, Tr. 13566.1
-599-
• :»
L
I j
A
■ !
.■i
t
: !
"3. Regular field inputs to data bases are clearly
modulated by quota objectives bias.
"4 . Recourse to the field for instant surveys
leaves them short, on time, us long on dependence — in
our survey we check out at about 85% accuracy.
"Conclusions? Forecast based upon CS Data Bases and
field surveys inherit a builtin error factor of +> 20%."
(DX 873, p. 33.)
Inadequate Financial Controls. A study of the Computer
Systems Division in the summer of 1971 reported:
"It has become apparent that CSD has not had adequate
financial controls and analytical capability. Because
of the complexity of the computer business in terms of
revenue and cost forecasting, the interaction between
generations of equipment, and the requirement for
large, direct sales force, the control and analytical
needs are greater in CSD than in most other businesses.
These controls and analytical skills have clearly not
been adequate in the division in the past. Moreover,
despite some awareness of problems developing, Corpo-
rate Finance did not provide the required support or
leadership to the division in up-grading the controls
and basic capability. If some of these problems had
been made clearer earlier, the business might have
been conducted in a different manner." (PX 349, pp. 6-7.)
The lack of financial controls resulted in RCA's
inventory being overvalued,* past due accounts receivables
* The 1971 Study of the Computer Systems Division reported:
"It has recently become apparent that a significant
portion of Computer Systems inventory may be overvalued.
Although it is not possible to identify at this time the
extent of the problem, major writedowns will be required
on video data terminals, Spectra 45 Mod 1, disc drives
and other computer equipment. Available reserves may not
be adequate." (PX 349, pp. 6-7.)
-600-
25
* The report from Arthur Young observed:
L i with significant amounts being prematurely written off,* and
2 ; questionable orders being booked.**
3 ! Product Deliveries . Because of product problems fore-
4-1 casts of RCA Series shipments were not met. For example:
5 [ "In the June 2 presentation, the 1971 business plan
[ assumed shipment of sixty RCA 6 series in 1971. As of mid-
S i July, Computer Systems indicated that the best estimate of
RCA 6 shipments for 1971 was fifteen. A similar decline
7 j has occurred in the case of the RCA 7.
3
9
10 1
"We believe that one of the most critical financial
11 j problems facing Computer Systems is with its accounts
receivable. Of approximately $27 million of billed
12 | receivables, close to 50% ($13 million) are past due, and
$3.3 million or 68% past due in Magnetic Products, Memory
13 i Products and Graphic Systems.
i
14. Ij "This situation has been caused by many factors. A
primary cause has been improper communication and incom-
15 j plete data flow between the field and Headquarters , and
within the Headquarters .
16 11
"Since many of the past due receivables are disputed
U | items, a significant portion of the receivables are being
written off or reversed, rather than cash being collected,
ig I Credits to receivables have been averaging $3 million a
jj month for the past twelve months ($37 million) . Receiv-
12 j able reversals in May totalled $4.4 million, $2 million of
;| which were netted directly against revenues. A limited
2q '{ two-week test of Task Force collection results on receiv-
•i ables past due over 90 days indicated that 40% were being
27 ij reversed and only 60% collected in cash." (DX 955, p.
~ ; 2; see also DX 11106, p. 1.)
22 H
23 ■
"A recently completed audit report indicates that
24 ! one-third of the bookings represent questionable items.
j ... This problem is compounded by the fact that a portion
of sales commissions are paid at the time an order is
booked and, therefore, salesmen may have been overpaid to
the extent that the bookings are not firm. (PX 349, pp. 6-7.)
-601-
** The 1971 Study of the Computer Systems Division reported:
"Despite assurances that time-sharing software
problems had been solved last fall, software avail-
ability continues to be a severe problem. ..."
(PX 349, pp. 6-7.)
<
RCA Series Impact. The NTS series appeared to be likely
to impact the RCA Series:
"Based on expected introduction dates for the NTS
series, it appears possible that a six-year life for
the RCA series will not be achieved. A shorter system
life would result in significantly greater write-downs
in 1971 and future years. This impact could be antici-
pated by increasing the obsolescence reserve or accel-
erating depreciation but either of these actions would
cause additional losses in the shorter term." (PX 349,
p. 7.)
Changes in Accounting Procedures . As noted above, premature
; returns of products placed under "accrued equity" contracts forced
RCA to take debits against current revenue. (Wright, Tr. 13589-91.)
A draft release of the Accounting Principles Board "put in question
the accounting practices being applied within RCA to the Accrued
i Equity lease".* (Conrad, Tr. 14057-58.) The effect of a retroactive
change in accounting practice would be large, involving a total effect
i of a $53.6 million reduction in revenue for 1971 and a $104 million
I
j reduction in 1972 with a "substantial negative effect on the P&L
4
i * This was the same ruling which affected Telex and Memorex and
r ! which would have required RCA to treat such transactions as leases
1 rather than sales, ceasing its practice of taking 70 percent of the
j, ! revenues to be received over five years as revenues received in the
j initial year of the contract. (Wright, Tr. 13590.)
-602-
i i performance of Computer Systems in 19 71 . . . and even
2. ; greater negative effect in 1972". (DX 956, pp. 5-6; see Conrad,
3; Tr. 14058-65, 14069-70.)
i
t
4-| The Economy. RCA's computer business was hurt by the
i
5 \ poor state of the economy in 1970 and 1971. "The economic
i
6 ! situation for the computer business in 1970 was quite bad. . . .
7 | Shipments that year were down some 20 percent from the pre-
i
S| vious year." (Rooney, Tr. 12264.) The economic situation
9 I increased the number of returns of computer equipment that
i
10, • RCA experienced. (Rooney, Tr. 12682.) The Returns Task
t
i
Xl ! Force estimated that 12 percent of the returns of the Spectra
\2 i in 1971 were due to the poor state of the economy. (DX 873,
i
13 i| p. 24.)
!
£4 ; Increased Competition. During the late 1960s and early
i
i
£5 i 1970s increased competition hurt RCA's computer business. While
i
t
IQ | RCA was putting out its old technology RCA Series , IBM was intro-
yj 1 ducing a series based on new technology. As discussed
above, prior to the announcement of that new IBM series
RCA had attempted to predict the price/performance of IBM's
anticipated new line in setting up its strategy. When the
2t ;J 370 systems were announced in mid 1970, RCA found that its
22 I predictions for the 370/155, 370/165 and 3330 disk drive were
2- l "accurate". Its predictions for the 370/135 and 370/145,
2* ! however, were "off target". RCA had anticipated that the
price of the 370/145 would be 5 to 10 percent higher than the
lS;i
19
20
25
-603-
RCA 6 price; as announced, however, in September 1970 the 145 was
priced approximately the same as the RCA 6. The 370/135 also came
out with better price/performance than RCA had anticipated. The
result of the inaccurate predictions were less sales for the Spectra
6 and 2 because those systems were not as price/performance
competitive as RCA believed they would be. (Rooney, Tr. 11910-11,
11913-14, 11922-23, 12583-90, 12649-50; PX 349, pp. 6-7.)
Moreover, Rooney testified that all of the significant
technological innovations in 1970 were achieved by the IBM
Corporation. (Tr. 12048-60.) Those included the 3330 disk drive,
which "brought to users significantly improved price/performance,
capability of storing and retrieving data on disks at much faster
speeds than [previously]" (Rooney, Tr. 12049); semiconductor
memory, which "reduce [d] the cost to the user in terms of the
amount of money he would have to pay for memory . . . [and gave]
the ability to have potentially much higher speed" (Rooney, Tr.
m ! j 12050-51, see Tr. 11923) ; microprogramming, which allowed the
t \
m ;| user to "improve the speed" with which he would process different
3
') applications and made it possible for comouters to more readily
:{ perform the instruction sets of other comouters (Roonev, Tr.
a i
I 12056-57); and the 3211 high-speed printer. (Rooney, Tr.
j 12058-59.) Rooney agreed each of the "significant innovations"
! attributable to IBM gave IBM "a competitive advantage in
3 1
>e
-604-
Ij marketing commercial data processing systems." (Tr. 12060.)
i
i ■
Z\ RCA's competition was not limited to IBM. Compe-
i
3 S tition from other sources was also increasing. A "Computer
i
4- 1 Industry Survey"- prepared by RCA in February 1970 listed 102
t
5 companies (other than IBM) offering "computers", "periph-
!
S; erals and components", and "software and services". (DX
i
7 j 1110 7, pp. 4-6.) RCA was experiencing increased competition from
3 | peripheral manufacturers. Wright, who was Chairman of the
i
9-j Peripheral Task Force in 1970, testified that the Task
i
IQ i Force was "surprised" and "shocked" by the number of users
H | using, or intending to acquire, non-RCA equipment as part of
12 \ RCA systems . This indicated to him that users "had learned that
£3 i it was possible for them to achieve certain benefits by pro-
|
74.';' curing and mixing boxes from different manufacturers in
* t
£5 i the same system". (Wright, Tr. 13555-57; see also DX 852,
i
Tg } PP- 14 > I" 7 / 19-20.) By July 1971 RCA's Data Processing
* i
yj 1 Division monthly report listed among "significant problem
areas":
IS
IS
20
"Independent peripheral manufacturers, i.e., Potter,
Singer, have been waging extensive sales campaigns
at selected customer sites. For example, Singer/
Frieden [sic] has proposed a plug to plug capability
2«r l! for replacement of the 70/564 Discs at California
~j Dept. of Justice." (DX 11099, p. 4; see also
~ n \ Beard, Tr. 9021.)
22 ;
I
22 j RCA also experienced increased competition from
-. ! third-party leasing companies and the same July monthly
-e ' report said that "discounts being offered on 360 systems
-605-
by third party leasing companies have [among other things]
accounted for the slowdown on the demand for RCA Series
systems".* (DX 11099, p. 5; see McCollister, Tr. 9290-92.)
d. RCA's Decision To Sell Its Computer Business To
S perry Rand. It was clear by the middle of 1971 that RCA's
computer business had been hurt substantially by management
errors, particularly by the introduction of the RCA Series.
Yet it was not clear that RCA needed, or even wished, to
sell its computer business. In 1971, according to Conrad,
RCA's management had "a very strong commitment" to its com-
puter business. Indeed, in July 1971, Conrad, speaking by
videotape to a Computer Systems Division marketing manage-
ment meeting, tried to dispel rumors that RCA would exit the
business. He spoke "in substance" as follows:
"As RCA's new President and Chief Operating Officer,
let me assure you that the goals of your division, as
enunciated so frequently in the past by Bob Sarnoff,
are not changed. They remain faithful to his often
expressed determination — to achieve for RCA a profit-
able, long-range computer operation.
"Every member of RCA's corporate management team
stands behind that commitment. Contrary to rumors, RCA
-\
1
21 il .
ij * McCollister testified that "the impetus for the concep-
22 f tion and the development of the use of" the accrued equity
ij contract, described above, "came about to some considerable
23 I measure because of the presence of leasing companies in the
j marketplace". (Tr. 9805.) Such contract began to be used
2d ; extensively toward the end of 1969 to the early part of
; 1970. (McCollister, Tr. 11038, see also Tr. 9804.) 3y
2c 1970 between 50% to 60% of the new contracts written were of
" : this tvpe. (McCollister, Tr. 9806-07.)
-606-
L
Z
z
4*
5
a
7
3
ia
Ll
IZ
13
14
IS ;
has no plans whatsoever to sell its computer operations.
As we informed the New York Stock Exchange less than
two weeks ago, these rumors are old hat. They've
been circulated in the past. They were unfounded then.
They are unfounded now*
"Neither rumors nor setbacks will undermine our
commitment ..to computers. We faced them before — as we did
in color television. We believe computers can do for
our company in the Seventies what color did in the Sixties.
"On a personal note, I had an opportunity last
week to express my personal commitment to RCA's com-
puter business. I heard that rumors about RCA selling
its computer business had caused the president of a
mid-western railroad to revoke his order for an RCA
computer. So I picked up the phone and called him
directly. I told him exactly what I have just told
you. And I've now been informed that the order has
been reinstated." (PX 192, pp. 3-4; see Conrad, Tr.
13939-40.)
He added: "We are making a greater investment in the com-
puter business than in any prior venture in our history.
This is a measure of our confidence that RCA systems and
._ [| products will effectively meet competitive challenges in
the decade ahead." (PX 192, p. 2.)
By September, however, RCA's view of its participa-
tion in the computer industry had changed.* A group of
executives consisting of Conrad, Sarnoff , Morsey and R. L.
21 I * In Conrad's "judgment" the internal discussions in
j RCA's management concerning staying or exiting from the
22 I computer systems business "really began in August of 1971".
j (Conrad, Tr. 13942.) On August 27, he and Sarnoff received
22 ■:. reports concerning the status of RCA's computer business.
| (See PX 201; PX 349.)
24. i
-607-
r
Y
)
I
4
Werner* (Conrad, Tr. 13942-44) met on September 16, 1971, for an
hour and a half and decided to recommend to the RCA Board of
Directors that RCA exit from the business. None of these four had
ever had direct responsibility for the RCA computer division or had
even worked in it. The Board of Directors adopted their recommenda-
tion on the following day.** (Conrad, Tr. 13942-43, 13948, 14145.)
The decision to sell the computer business came as a
"surprise" to persons working in the Computer Systems Division.
(Rooney, Tr. 12369; Wright, Tr. 13172, 13570-71.) RCA's management
had not consulted with Wright, Donegan (Vice President and General
Manager of the Division), Rooney or, to Rooney' s knowledge, anyone
else in the Computer Systems Division. (Rooney, Tr. 12368-71;
Wright, Tr. 13572.)
The basis for the Board's decision to sell its
business was that if anticipated losses of $137 to $187
^ ' million in the Computer Systems Division over the period
,7 ;:
3 •
j * Werner was RCA's general counsel. (PX 341, p. 38.) He retired
•9 ; j as general counsel during March 1978 and continued as a director j
;i until 1979, when he did not stand for reelection. (DX 13853, p. 41; i
& j DX 13902, p. 48.) Conrad left RCA in 1976 (DX 13852, p. 37), and j
i Morsey left in 1973. (DX 678, p. 40.) - j
•j ** At the directors' meeting the recommendation was embodied in a j
VL) memorandum (PX 208, p. 8), which Morsey read to the Board of i
I Directors. (Conrad, Tr. 14072.)
!
.1
It !
-608-
L
2
3
4-
f
7
a
3
ia
u
1971-1976 materialized, that would cause a need for a
greater investment in computers than RCA chose to make. The
amount of the investment required over the 1971-1976 period
was estimated to be $702 million. (PX 208 , pp. 4-5* ) The
amount was disputed, however, even within RCA, and some
j thought the figure was overstated by $100 to $200 million.*
Competing with the computer division for invest-
ment funds were the many other divisions in the RCA Corpo-
ration. The needs of those other divisions for investment
funds also were greater than had been expected; a "pre-
liminary evaluation" showed that "new funds required" during
1971-1976 "may exceed $1 billion" for the corporation as a
i
13
»
y, l * Julius Koppelman, the financial vice president of the
Computer Systems Group, believed that the $70 2 million
figure was in error by a very large amount. He, of course,
had not been consulted before September 17. (Rooney, Tr.
.„. 12370-71.) Wright testified that:
la :t
; "Julius Koppelman . . . subsequently had occasion to
k * ij look at those financial numbers that had been put
together by the corporate staff and submitted to the
board, and Mr. Koppelmann told me that there had been a
■j serious error in those calculations on the order of a
£- | hundred million dollars, where it was overstated as
,| far as the amount of cash that was required, and that
22 j this had been caused primarily as a result of the
| corporate financial people not understanding certain of
21 j - the contractual terms and conditions that we had with
A customers on certain of the equipment that we had
22] installed." (Tr. 13572-73.)
23 j Similarly, Rooney testified that he was told by Donegan
j "that Mr. Koppelmann . . . told him that there was an error
2A ) of some $200 million in those numbers ... in overesti-
! mating the capital requirements of the business." (Tr.
25 j 12371-74.)
-609-
whole. (PX 208, p. 5.)
It was against that background that RCA made its
decision. It considered whether to proceed with the magnitude
of investments contemplated both in computers and other
areas, and believed that "if earnings growth can be maintained
!i at an annual rate of 10%-15%, the Company can raise needed
funds". However, if RCA earnings were only to grow "at a
rate similar to GNP (7%)" or if a recession were to occur
"the resulting reduction in RCA's overall profit position
could bring considerable pressure on obtaining the $1 billion
outside financing required". There could be "severe financing
problems". Major losses in computers would add to the
difficulties. (PX 208, p. 6.)
Conrad testified that he believed that RCA could
have raised the necessary capital to finance the projects
that the Computer Systems Division had in mind at the time,
■[ could have reached its goal of achieving 10% of some
'•I
I defined market, and would at some point in time have been
l| profitable. (Conrad, Tr. 14047-48, see also Tr. 13944-47.)
j Similarly, Rooney — who as a member of the Computer
T !
•| Systems Division had not been consulted prior to the decision —
1 'j ■
■j testified that he believed that RCA could have been successful
2 ••;
| in displacing IBM products in the 1970s had it been allowed
3 i
to continue in the business. (Tr. 12094.)
A i
j Withington, who had advised RCA's Computer Systems
-610-
I
:
3
7
I
3 I
L| Division a year earlier, believed that RCA could have been
2 ; successful in the computer business had it chosen a differ-
31 1 ent, less adventurous, strategy and remained profitable while
4-|* growing more slowly. (Withington, Tr. 56711, 56720-21.)
i ■■
5* i ; RCA considered what actions could be taken to
><
I'
S. J! mitigate the risk that it would need a large amount of
7 j| financing if it stayed on its present strategy in computers.
3 I One alternative mentioned was to
g; |; "reduce investments in other new businesses such as
i the Supermarket project, the French Color Tube venture,
£Q •' SelectaVision, etc. , thereby substantially reducing
i. outside financing requirements. Even in this case, if
j^ (■ profits were to grow at 7%, or if a recession were to
I occur, it is likely that the major financing which would
n \ still be required for CSD could only be obtained at
| higher interest rates and more restrictive terms . In
£3 ; addition, RCA would in this case be relying even more
directly on the ultimate success of CSD by passing up
T( l '} other, perhaps more attractive opportunities." (PX 208,
p. 6.)
is ;i
17
Another alternative would be to keep the computer
business but "to substantially reduce the size and scope of
the computer operation" by limiting it to certain narrow
-^ it market areas. However, it was believed that this would
tc 1
iS ;} reduce revenue as well as expenses and "while cash requirements
would be reduced substantially, it is questionable whether
the business would ever attain economic viability". ( Id. ,
p. 6.*) As a result, "the additional' investment required in
& ■!
H .1
L.Z. ■(
4A \
!
i
\
24 1
* Other alternatives mentioned were to cut dividends and sell
surplus real estate and "marginal business". (PX 203, p. 6.)
-611-
\
}
1
I
CSD no longer appears to be a prudent financial risk" .
( Id. , p. 7.)
A major concern was that if there were a downturn
in the economy in the mid-1970s r "the many healthy and vital
parts of the rest of RCA could be hindered" because of the
"high level of outside financing" required for computers
"as well as the other parts of RCA" . "While the computer
industry is an attractive and growing business — although at
a slightly lower level than originally anticipated — the
profits to be gained by RCA from this business, in relation
to the total investment required, do not appear to be con-
sistent with sound financial planning" . ( Id. )
In addition, the memorandum said that given RCA's
\ position in the computer business "[t]he manpower and finan-
cial resources of IBM, including the size and strength of
the marketing, research and development organizations, are
such that achieving market share growth as well as accept-
able profitability, is extremely difficult". (PX 208,
p. 7.) Thus:
4-
:t
si
»
i
"In summary, the computer business currently
accounts for about 6 percent of RCA's total revenues.
While it could represent a growing segment of the
LI | Company's operations, it is unlikely to ever exceed
■j perhaps 10-15 percent of total RCA volume. Continued
:2 ;l commitment to computers, however, could lead to severe
j financing problems for the Company and may contribute
23 | to restricted growth in other operations. On balance,
! it is believed that the risk does not justify the
l*. j potential reward. Therefore, withdrawal from the
j mainframe computer and peripheral equipment business is
?« I recommended." (PX 208, p. 8.)
-612-
2;
i
^ i
1
i
_ !
3 i
!
5 j
3 |
9 I
1C j
11 !
i
12.
13
15
16
17
IS
Similarly, Conrad testified that RCA left the
computer business because:
"[Wje believed that we could apply the resources
employed in the computer business better in other
opportunities within RCA, which would lead to
profit at an earlier time with greater assurances."
(Conrad, Tr. 13933-34; see also. PX 217, pp. 2-5;
Conrad, Tr. 13989-94.)*
All the participants and observers of RCA's deci-
sion to sell its computer division agreed that RCA's deci-
sion was voluntary, and "nothing IBM did or any other
company forced us or caused us to exit the business per se" .
(Conrad, Tr. 14046; see also PX 217, p. 5; Rooney, Tr. 12387;
Wright, Tr. 13630-31.)
After September 17, 1971, RCA negotiated with
Sperry Rand and with Mohawk Data Sciences concerning the sale
of its computer business, as well as having meetings with
several other companies , including Burroughs , Xerox and
Memorex. It sold the division to Sperry Rand.** (Conrad,
Tr. 13953-54, 13968-70; PX 402.) In its presentations to
•t
:l
Q
ij * Among the projects in which RCA believed it could
2G j invest its money with more opportunity for profit than the
i computer business, the principal ones were Alaska Communi-
21 ; | cations, the Global Communications Company, SelectaVision,
•land VideoDisc. (Conrad, Tr. 14046-48.) By 1978 RCA had sold
22 j Alaska Communications. (DX 13854, p. 15..)
22 ;| ** During the period subsequent to 196 6 RCA also withdrew from
many businesses in addition to the business of its Computer
4* Systems Division. Those businesses include: the manufac-
turing and marketing of radios, phonographs and stereo equip-
25 ment; the manufacturing of radio tubes; a color picture tube
-613-
prospective purchasers, RCA estimated that the "After Tax
Cash Contribution" of its lease base for the period 1972
through 1974 would be $193 million assuming no residual
value. (PX 405A, p. 8.) It sold its computer division
to Sperry for approximately $137 million. (Conrad, Tr. 14130.)
RCA reported that it lost approximately $241 mil-
lion before taxes on its computer systems operations for the
years 1958-1971*. (PX 410.)
f. j In September 1971 RCA set up a reserve of $490 mil-
i
j ; lion pretax, $250 million after tax, to cover prospective
losses in connection with the sale of its computer division.
The losses that were anticipated related to disposition of
• i assets "such as inventory, receivables, plant", "discharge
of claims and obligations for commitments to employees for
>«• !
:. joint venture with Thorn Electrical Industries; Meyer Bros.
i Parking Services; United Exposition Services (a subsidiary
' ; of Hertz that engaged in services related to exhibitions) ;
:| Cushman & Wakefield; the design, manufacture and sale of
* ',i microwave communications transmitters, receivers and multiplex
:{ equipment in the United States; electron microscopes;
- :i Service America, which offered to service televisions of all
:j manufacturers; RCA Alascom; and Random House. By 1980 Banquet
2 ,j Foods was also up for sale. (Conrad, Tr. 14022-27; DX 13854,
t j p. 15; DX 13860, p. 8; DX 13902, pp. 2, 36-37.)
\ * RCA's losses in the late sixties had "to do with the
— .! investment that we felt we had to make and the engineering
;j and programming for future profitability in order to grow
2 | in *he business", as well as to RCA's accounting for leases.
" [Pjrofitability was governed primarily by the rate at which
RCA determined that it would like to grow", about 20% a
year, "somewhat faster than the general growth of the market".
5 ! (Beard, Tr. 8535-38.
-614-
L |
2 |
i
4-i
3
9
II
12
13
L4
16
i
IS
19 :l
severance and release", and other purposes. In December
1973 a review- indicated that the disposition was going better
than expected and the reserve was reduced by $78 million,
leaving a pretax reserve of $412 million. (Smith, Tr. 14247-48.)
e. After the Sale to Sperry Rand. The story of
RCA's participation in the computer industry after the 1971
sale to Sperry Rand has two parts: RCA's activities and
Sperry Rand's success with the computer division it purchased
from RCA.
(i) RCA's Activities. As it had planned, RCA invested
heavily in other businesses in the 1970s. Conrad testified
that from 1972 through the end of 1976, RCA invested approxi-
mately $130 million in satellite communications (Tr. 14009-10) ,
more than $250 million in the same period in Global Communica-
tions (Tr. 14011-12) , approximately $200 million annually in
the purchase of automobiles for Hertz (Tr. 14102-06) , and $150
million in Alaska Communications. (Tr. 14008-09.)
j RCA also continued in or entered computer-related
i
i businesses. Conrad testified that:
j "In our Solid State Division, we manufacture, design,
^ j engineer and manufacture integrated circuit devices
called microprocessor chips, which can be and are
used in data processing applications, as well as com-
munications applications.
21 •!
•4
"We also continue to offer and perform service
22 j on a variety of data processing or reservation system
j terminals, which are owned by others.
t
J "We continue to from time to time design, develop
-615-
and manufacture special processors which are sold to
the government in conjunction with tracking devices,
such as radar." (Tr. 14048-49, see also Tr. 14157-58.)
(ii) Sperry Rand's Success with RCA's Computer Systems
Division. Sperry believed that its acquisition of RCA's
Computer Systems Division was "a sound business" and a
"wise" decision. (McDonald, Tr. 3873-74; DX 63, p. 1; DX. 71,
p. 9.) In its 1973 Annual Report Sperry reported:
"More than 90% of these RCA customers remained with
us, and more than $130 million in new equipment was
shipped to these users during calendar year 1972.
We are continuing to build ' bridges ' between the RCA
systems and Sperry Univac's line, and we are confi-
dent that many of these customers will eventually
convert to Sperry Univac's systems." (DX 63, p. 1.)
In December 1974, 77% of the original RCA customers acquired
by Univac were still using their RCA equipment and 5% of
L ;j the original customers had moved to Univac systems , and the
RCA equipment had yielded a "revenue stream (sales, rentals
and maintenance) for 3 years of approximately $370 million".
Univac believed' that "these benefits will certainly not end
at this point". (DX 68, pp. 11-12.) By May 1975, approxi-
mately 76% of the RCA equipment acquired by Univac was still
on rent. (McDonald, Tr. 4045-46.)
f. Conclusion. Like General Electric, RCA was a
k '
7
3
7
3
T
} large company with a small computer business. In the
2 -|
last full year before its sale to Sperry Rand, RCA's U.S
1
EDP revenue was $226 million. (DX 8224, p. 2.) RCA's
venture into computers was a failure; but it need not have
-616-
L j been. As we have seen, despite RCA's great technological
2. | capability in the 1950s, RCA only placed nine computers in
that decade. RCA's inactivity in the 1950s and early 1960s
4*jf cost them, dearly. But RCA still could have succeeded. The
5 [ : Spectra, patterned after IBM's 360 was a mixed success.
I
a i Reliability problems and inadequate peripherals limited the
7 ;' acceptance of the systems. But even then had RCA understood
g || the need to push ahead with technological development, to
g. I commit its ample resources to new, more advanced follow-on
iq ! systems, it could have succeeded.
i
££ ! Instead, it introduced the RCA Series — yesterday's
£2 j technology at lower prices — and it chose that vehicle to
!
£3 I spearhead its drive to "gain market share" and "become number
T< # : 2 in the industry". But the RCA Series could not compete
£5 1! with the more advanced products of IBM and others and was a
"major product failure", blowing the Spectra series "out of
the water" .
At the same time, the management of RCA changed
hands and the company sought to transform itself into a
conglomerate. The result of its conglomeration was that
all the various corporate mouths needed feeding at once
16 il
i
is;
is!
20]
?7
•i and as the company entered the recession of the early 1970s,
it found itself stretched too thin to pay adequate attention
23
4A I
-I
t
i
or commit sufficient resources to save the computer business
from the RCA Series debacle.
-617-
r j
i
I
I-
:
m
3
A
.1
°
? :l
2 \
W- i
i
IS i
In sum f the story of RCA, like the story of
General Electric, is the story of missed opportunity, bad
management and product failures.
•\ !
-618-
2
5 |
7
S
43. Honeywell. The history of Honeywell during the period
1964-1970 was built on the success of the Honeywell 200 — a product
which gave birth to a compatible family of computer systems and in
r turn sparked expansion of Honeywell's peripheral line and service
capabilities. Despite some difficulties along the way, Honeywell
ended the sixties with a large and successful array of electronic
data processing products and services with rising revenues and profits
derived from them.
a. The 200 Series. In December 1963 Honeywell had announced
n !J its 200 computer system, along with an "automatic program conversion
n | package, called 'Liberator'". (DX 198, pp. 25-26.) Richard Bloch,
: who led the Honeywell team that designed the 200 (Tr. 7886), testified
: to the strategy behind it:*
i
14;
j ; * Richard Bloch, who was Vice President for Product Planning at
^3 j Honeywell for most of the period he was there, 1955 through 1967,
f testified about Honeywell. His duties as Vice President were to
16 ;l develop product, pricing and marketing strategies for the products
•of the EDP Division. (Tr. 7575-76.)
17 I
: James H. Binger and Clarence W. Spangle were the other witnesses
IS 'I who testified about Honeywell. James H. Binger was Chairman of the
•| Executive Committee of Honeywell, Inc., when called to testify. (Tr.
19 j 4489 . ) An employee of Honeywell since 1943, he became chief executive j
'} officer of Honeywell in 196 4 and held that post for the following j
20 1 decade; starting in 196 5 he was also Chairman of the Board of Directors. j
| (Tr. 4489-90.) |
21 it i
\ i
4 Clarence W. Spangle was President of Honeywell Information Systems j
22. : | and Executive Vice President of Honeywell', Inc. in 1975. (Tr. 48 82.) j
| Spangle, too, had been with Honeywell since the 1940s. ( Id. ). From I
23 1965 through 1969, Mr. Spangle was Vice President and General Manager j
i of Honeywell's Electronic Data Processing Division, responsible for j
24. [marketing, manufacturing and development of data processing systems. !
.} (Tr. 4887-89.)
-619-
L
2
3
t
i
Si
i
i
7 j
i
3 i
i
i
I
ia :
i
n !
12
13
14*
15
16
"[T]he 200 was conceived to represent a next step for 1400
Series users in the IBM line, and we really designed that
machine from the outset to be attractive to that user community.
"One of the attractive features had to be a means of
getting the user to bri,ng his programs on to the new machine,
and that required conversion facilities and conversion
offerings, which we intended and indeed did develop." (Tr.
7578.)
Honeywell felt this strategy would give it an "accelerated move into
the [general purpose business data processing] field, which we needed."
(R. Bloch, Tr. 7585-86.) In 1963 Honeywell's EDP revenues were only
4% of its total revenues, and it had yet to make a profit in that
area. (DX 132, p. 11; DX 198, pp. 4-5; DX 8224, p. 387; DX. 14484,
p. Rl; DX 8631, pp. 31, 37.)
The 200 was designed to make conversion from the IBM 1400
series as easy as possible. (Binger, Tr. 4823; R. Bloch, Tr. 7886.)
An effort was made to replicate closely the file structure, media and
formatting of the 1400 (R. Bloch, Tr. 7605-06; Spangle, Tr. 5025) and
the LIBERATOR conversion aid was developed. (R. Bloch, Tr. 7606.)
U ;j The LIBERATOR enabled 1401 programs to be converted to 200 series
13 i programs by means of assembly language and object code translators,
ig 'J ( Id. ; Goetz , Tr. 17652-53.) Because the conversion required only
20 ] a very small amount of manual intervention, it resulted in a high
21 jj degree of efficiency. (Spangle, Tr. 5021-22; Goetz, Tr. 17652; see
22 ; ialso R. Bloch, Tr. 7888-89.)
22 ! ! The LIBERATOR successfully accomplished the conversion for
■I !
2* : |which it was designed. (Binger, Tr. 4823; R. Bloch, Tr. 7888-89; see i
■i !
?c '-also Goetz, Tr. 18780-81.) Thus, the 200 offered users both an easv |
I I
-620- !
L
Z
3
7:
3
9
ia
u
12.
conversion method* and price/performance superior to its competitors,
including the 1401. (McCollister, Tr. 11237, 11365-66; Evans, Tr.
101187-88; PX 6204, p. 4; DX 167.) Both of these characteristics led
■ the Honeywell 200 to enormous success. (R. Bloch, Tr. 7602-03, 7888;
McCollister, Tr. 11235-36; Withington, Tr. 55863-67; J. Jones, Tr.
78989-95.) Withington wrote in October 1964 that Honeywell had
obtained many hundreds of orders 'for the 200, and that no computer
manufacturer was gaining ground as fast as Honeywell. (PX 4829, pp.
20-21.) IBM reports called the 200 an "outstanding success",
allowing Honeywell to expand its marketing and other personnel and
to turn a profit for its ZDP Division. (PX 3481, p. 69.)
McCollister testified that Honeywell expanded its sales force during
the early sixties so that by 1965 Honeywell had 50 to 75 percent
. , ! l more people than did RCA, although the two companies had started the
I sixties equal. (Tr. 10962-63.) Sales were made to all kinds of
customers; Gordon Brown testified that as a Honeywell marketing
; representative he sold 200s to an insurance company, an aircraft
1 company and a service bureau; Honeywell reported sales to the Internal
12
13
16
17
IS
it
; j Revenue Service and to U.S. Air Force Major Air Commands. (G. Brown,
Tr. 50985; DX 13349, d. 27.)
2Q .1
i Within IBM, the Honeywell 200 announcement provoked heated
21 J .
i discussion on how soon IBM was going to .come up with a better per-
22 ;;
| forming product with which to respond. Immediately following the
23 1
25 .1 * The 200 was also compatible with "most widely used
| small computers". (DX 167; DX 198, p. 26.)
-621-
1- i announcement, T. V. Learson, IBM's Senior Vice President, wrote to T.
2 ; J. Watson, Jr. , its Chairman, and A. L. Williams, its President, that
3" ! the Honeywell announcement was "even more difficult than we antici-
4*1 pated". (PX 1079.) Shortly thereafter, M.. T. Hague, IBM Director of
i
5 I : Market Programs, wrote to Dr. J, W. Gibson, IBM Vice-President and
5-i Group Executive, that the 200 "represents the most severe threat to
i
7 j IBM in our history". (PX 3912.) Evans testified that the marketing
i
!
8 'force and others dealing with the 200 regarded it as a real challenge
9; to IBM (Tr. 101186), and both he and Knaplund testified that the
IQ : marketing organization put a lot of pressure on the development
I
11 j organization to announce the 360/30 as soon as possible in reaction to
i
12 I the 200. (Knaplund, Tr. 90475; Evans, Tr. 101190.) On January 28,
13 j W. C. Hume, President of the Data Processing Division, wrote to Dr.
i
14. ; Gibson:
t
15 i "We must have an answer to this system immediately ....
| The best solution to this problem ... is a 101-H [360/30]
15 ! machine with a competitive price to the H-200 and a
ii performance equal to or greater than the H-200, ready for
17 I announcement by mid-February." (PX 1090; see pp. 353-57 above.) "
13 ; Of course, by April 1964 IBM was able to respond to the
iq | H-200 with the 360/30 — which was two or three times more powerful
I
2Q j than IBM's 1401 at less than one and a half times the price. (Hughes,
Tr. 33924; JX 38, p. 33; DX 573, p. 6; see pp. 230-31 above.) Despite
21
22 •; the announcement, Honeywell continued its successful course. In
22 ; ! December 1964 T. V. Learson wrote to T. J. Watson, Jr. that "the
25 j
i
1!
-622-
L ; Honeywell 200 story" had led to 300 losses to date for tape-oriented
2 ! systems, with 1,000 such situations in the doubtful category, 40% of
3 !. which he estimated as losses. (PX 1288, p. 2.)
4» V Honeywell spent the remainder of the decade enlarging on
it
5" if and solidifying the 200 's success. In a 1969 speech, Binger outlined
his strategy:
3
T
3
10
11
12
13
L4-
16
17
IS
"In the beginning we made a conscious decision and adopted a
strategy to compete in a broad segment of the computer
marketplace, and to make significant penetration through a
wide array of products and services . Our highly successful
Series 200 computer line is the prime example of this
strategy." (DX 132, p. 12.)
In June 1964, after IBM's announcement of System/360,
Honeywell announced the 2200 and followed it in February 1965 with
three other compatible new members of the Series 200, giving it a
"family of computer systems": the 120, the 1200, the 2200 and the
4200. Honeywell stated that "[t]he family concept of these new
systems gives our customers the assurance that they can meet
j problems of growth by expanding through an extended range of central
| processors , continuing to use the peripheral equipment already in
i their EDP system."* (DX 13849, p. 27.) The 200 series
** | was also, through hardware design and programming adaptations,
22 ,j "accessible to [Honeywell's] 400 and 800 [users] who can shift to
~ 1 the higher levels of the newer series with a minimum of adjustment,
■•i
I and with the protection of a substantial part of their prior
23
Z- \ * The compatibility of the 200 family meant each model was also
J compatible with the IBM 1401. (Withington, Tr. 56375-76.)
7«
-623-
r
1 \
i
i
3 i
L
1
2 »
4-
6 ,
7 i
S
9
:c
programming and file investment." (DX 199, p. 32.)
Spangle testified that the 200 line was priced so that the
three-year lease prices would be "roughly equal to those of IBM for
equivalent price/performance on a system basis", with the one-year
r if price "slightly above that of IBM" and the five-year price "5 to 10
percent below the one-year price of IBM" . The consideration of what
the price should be was based on the price for the system, although
the individual elements of the system were priced individually.
(Spangle; Tr. 5056.)
According to Richard Bloch, Honeywell gauged its pricing
"against the nearest competitive IBM line or the IBM equipment which
we were hoping to supersede, to a lesser extent some of the other
competition" because "if we were to increase our penetration of the
market we would obviously have to take away some of the captive
business that was presently in IBM's hands". (Tr. 7596-97.) Honey-
well priced its product so that it demonstrated performance advantage
over IBM "that might be measurable in tens of percent" with a price
equal to or less than the IBM price. Where, on the other hand,
i
'j Honeywell felt that it did not have any substantial performance
:| advantage, it considered that it would certainly have to have a
i significant price advantage* This meant 90% or less than the IBM
"" '•• i
.j price. However, there was no "automatic rule of thumb". (R. Bloch, s
12 'I j
I Tr. 7599-601.) I
S |
As IBM improved the capabilities of its 360 line (see pp. ' :
U ' J
f above) , so did Honeywell with further improvement of the capabilities 02
15 "j !
] i
1 -624- I
: ! I
i • !
the 200 family through peripheral and software announcements. A number
of new products , both hardware and software , were announced for the
Series 200 line at the end of 1966, covering mass storage, data
communications and expanded multiprogramming, including four magnetic
disk devices for "random access information storage and retrieval"*
and a number of terminal devices. (DX 199, p. 31.)
The tempo of announcements accelerated as the decade went
on. In its 1967 Annual Report, Honeywell stated:
"Product lines of the Electronic Data Processing and
Computer Control Divisions [**] are being broadened con-
tinuously to assure competitiveness. In 1967, for example,
more than a hundred hardware and software products and
product modifications were added to the Series 200 EDP
line and the control computer line." (DX 200, p. 31.)
Binger and Bloch testified that Honeywell's competitors in
the 1960s were IBM, Sperry Rand, RCA, Burroughs, GE, CDC, NCR and
DEC. (Binger, Tr. 4527, 4593-94; R. Bloch, Tr. 7592-94; see also
Spangle, Tr. 4933-34.) SDS was a competitor "to a limited extent".
(Binger, Tr. 4515.) Honeywell also faced efforts by Fujitsu, Philips
and Nixdorf to sell their computer equipment in the United States .
* II (Binger, Tr. 4516-17.)/
J !
I
;j * These were acquired OEM from CDC. (See below, p. 628.)
- is
j ** See below, p. 633.
/ Other foreign competition was encountered, too; for example, Honeywe 1 ]
bid against Siemens, among others, in 1968, for an accounting and j
, payroll system for the U.S. Army. (DX 7556.) j
-625-
i :
\
l
I
l !
1
t
S
7
S
si
•l
Honeywell's 200 series was sufficiently popular to be
marketed by leasing companies. Thus, Leasco dealt in Honeywell
equipment before 1967 (Spain, Tr. 88749) and Finalco was leasing
Honeywell 200s and 1200s, a fact which made Patton, a Honeywell
Regional Sales Manager, "a little nervous over what could happen if"
those systems come off lease". (Spangle, Tr. 5191-92; DX 161A-)
Transamerica also leased Honeywell equipment in the late 1960s.
(Spangle, Tr. 5190.)
b* Problems and Solutions.
(i) Other Systems . Honeywell was not without problems over
this period, though. One of these was the 8200 computer system,
which was planned to be the most powerful computer system in the
200 series. (DX 13849* p. 27.) The Honeywell 8200 was announced
in 1965. ( Id. ) It was intended to bring together the Honeywell
200 line and the Honeywell 800 line, the latter of which was in-
stalled at that time at about a hundred different sites. (Spangle,
Tr. 4997.)
"At the time of the announcement the development of the
machine had not begun. And as the development was undertaken, it
■) turned out to be much more difficult to do those things than had
a .!
■j been anticipated," and Honeywell spent large amounts of money, more
'1 } .
j than it had planned, to develop the equipment and to develop the
j software to supply with the equipment. ( Id. )
j Honeywell was not able to achieve the objective of having
?.4 'i
-"* I
.{ that system be an upgrade path for the 200 line so, according to
Spangle, "its market became limited really to those 800 customers
-626-
L !
i
2 '
i
i
i
* I
I
5 I
7
3
9
10
11
12
13
14
15
16
17
IS
who wanted to continue largely in the batch processing mode and
wanted higher throughput". (Spangle, Tr. 4997-98.)
As a result of all this, Honeywell was able to ship only
about 40 of these machines which Spangle testified was not enough to
make the whole investment and development worthwhile. The particular
problem that caused the 8200 to fall short of its objective was the
need for two operating systems in one computer system — Honeywell
could not get it to work. (Spangle, Tr* 4998.)
Honeywell tried to aid its customers with 800 systems
installed in another way — by the provision of a larger system which
was compatible only with the 800. Thus, it announced the. 1800 in
1962. (See p. 189 above.) However, according to Withington,
the 1800, designed to accommodate the growth needs of the 800
users, "sold in only very small amounts". He attributed this to the
fact that the IBM 360 and GE 600 series, available at the same time,
"were regarded as superior to the Honeywell 1800 by users and
Honeywell users who outgrew their Honeywell 800 apparently more
frequently left Honeywell for a competitor than accepted the 1800
T q If instead" . The 360 and GE 600 systems were felt to be superior to
3g ; | the Honeywell 1800 "because they offered early versions of operating
2i :! systems whose primary initial virtue was to permit multiple pro-
22 I' gramming . . . plus automatic control of peripheral equipment in
~~ 'I ways which would simplify the users' programming requirements".
: i
- A J (Tr. 56491-93.)
;l
25 J
II
;i -627-
(ii) Peripherals . In the early sixties, Honeywell still
believed that magnetic card devices would be competitively superior
to magnetic disk drives. It had under development magnetic card
devices which had been announced to customers. However, the slow
speed and unreliability of the card devices caused difficulties and
hurt Honeywell in its marketing of systems. (Withington, Tr. 56494-
95.) Finally, the effort was dropped, termed in IBM reports "a
dismal failure". (PX 3481, pp. 75-76.)
Honeywell made its decision to abandon the magnetic card
mass storage devices following IBM's announcement of the 2311 disk
drive for the System/360. Withington testified that this was "a
major change for Honeywell, because at the time there was no expend! -
^ '* ture whatever for disk drive development, all of the mass storage
<3
.3
J
development efforts being put into the magnetic card devices, so
Honeywell had to start a new effort from scratch and also search
the industry for OEM sources for suitable disk drives". (Withington,
Tr. 58562-63.) By 1967 CDC was shipping its 9492 disk file to
L9
20 '!
| Honeywell, who then became its principal customer for CDC 9433 and
.s : ;
9434 disk drives, taking in excess of 4,700 units. (G. Brown, Tr.
51033-34, 51056-57.) Honeywell began to manufacture its own disk
! packs in 1967, but continued to purchase the drives. (G. Brown,
21 :i -
•j Tr. 51056-57; DX 200, p. 31.)
22 ; |
Honeywell had already begun efforts to produce itself all
j of the peripheral devices contained in its EDP systems in the mid-
24 1
•j sixties. (DX 13849, p. 28.) During 1965 it started deliveries
25 i
■! -628-
L
2
3
4-
5'
a
of its own card reader, and was about to start shipment of its
own card punches, (DX 13 349, p. 28.) Prior to 1965 Honeywell
purchased IBM card, readers and card punches and offered them with
its own computer systems, including the 200. It planned, however,
in that time period r to develop its own manufacturing capa-
bility in punch card equipment, a decision which was accelerated
7 if by the announcement in late 1964 that IBM would no longer lease
such equipment to Honeywell and other manufacturers planning
to re-lease them to customers but would only sell. (Binger, Tr.~
4512-13, 4549-50; Spangle, Tr. 5102-07.)
During 1965 Honeywell introduced new models of printers
and tape transports and started deliveries of a variety of communica-
tions terminals as well. (DX 13849, p. 28.) However, it
continued to acquire software or software development from
outside companies. (Spangle, Tr. 5092-94.) Contrary to
IBM's full-scale entry into the manufacture of its own
components in 1961 (see pp. 282-90 above), Honeywell divested
itself of its component operations in 1965, stating it "felt
that we should concentrate our attention on electronic end
: f products rather than components of this type. We intend to
20 \
i rely on the numerous well qualified suppliers of semiconductor
21 I .
% devices for our substantial requirements." (DX 13849, p. 6.)*
22 j
3
9
ia
13
14-
13
IS
17
IS
19 1
23
| * Honeywell reentered the development and manufacture of
24. 1 componentry in 1969 with the announcement of a new integrated
j circuit development center. (DX 123, p. 41.)
■i -629-
c. Marketing Practices . Starting in 1965 Honeywell
offered three- and five— year lease plans for systems and peripherals .
(Spangle, Tr. 4953;: Brown, Tr. 52613; Withington, Tr. 56624.)
Mf Under Honeywell's five-year lease plan, the customer could change
f I his configuration within certain limits without penalty. (Spangle,
i! Tr. 5529-30.) According to Honeywell's 1965 Annual Report, "the
j
T j extended lease terms for our new generation equipment have been
I j well received by our customers with the result that a substantial
a. ; percentage of the present leases are being lengthened to five
i ■
j | years. Use of extended lease terms by both present and future
j
^l- customers permits computer users to benefit from the new level of
i
?\~ technological stability which now exists". (DX 13349 , p. 5.) In
" !
!
i
I ';■ 1967 nearly 70 percent of "commercial" Series 200 contracts signed
i
^j were for five-year periods. (DX 200, p. 32.)
Bloch testified that during his tenure at Honeywell,
c \ (1955-1967) , 80 to 90 percent of Honeywell's computer systems were
leased. This was dictated "pretty much" by the customer. (Tr. 7673,
7675-76.) Honeywell developed a sale and leaseback method of
financing these leases in 1966. Honeywell would arrange with some
lending organizations — "usually these were syndicates put together by
someone; like say, the First National City Bank did one group"--
whereby it would sell them an amount of installed equipment and then
lease it back from them. They would give Honeywell the cash for
the equipment and then it would pay them in installments until it
had paid back the amount of the cash advance plus a financing
' ;I
si
«i
i
'1 ]
■i
i
i
-630-
t
al
!
charge. (Spangle, Tr. 5076-77.) According to Spangle, this "improved
our profit and loss statement" and produced "more cash with which to
operate". (Id.)
4*- f ' The sale and leaseback method continued until 1967, when a
5| wholly-owned subsidiary called Honeywell Finance was set up. That
S if subsidiary "was able to accomplish much of the benefits of the sales
7 i and leaseback transaction in so far as creating or attracting cash
3. I and capital . ► . although it did not have the effect of accelerating
9- ! the profit from the lease part of it as the sale and leaseback
IQ i transaction did". It did, however, preserve the residual value of
I
H | the equipment for Honeywell. ' (Spangle, Tr. 5082.) Honeywell Finance
X2. 1 borrowed money from banks and investors through commercial paper
13 I issuance and through the issuance of long term and medium term
y±\ bonds, on the security of the receivables from Honeywell's rental
jjj < contracts. The loan proceeds were passed through to Honeywell.
i
IQ ;| (Spangle, Tr. 5082-84.) Honeywell's initial investment in this
yj ! subsidiary was $15,000,000, half in a subordinated loan and half in
|q :j common stock. A $60,000,000 line of bank credit was established of
12 ; which $23,350,000 was being utilized at year end 1967. (DX 200, p.
20 ! 9 **
2t ij During the middle and late 1960s, Honeywell supplied
22 '■ educational courses for customers, programming support, operating
I
23 i| systems and application software without separate charge. (Spangle,
!
. A -I Tr. 5084-86.) Bloch testified that this was due to "the dictates of
:1
mm ' the marketplace . . . the traditional way in which these services
63 ii
j -631-
;!
:l
'i
•I
Ij
i
a;
3!
I
i
i
i
7 \
3 !
I
i
10 1
I
11 !
12 J
13 1
14,1
i
15 !
!
16 !
17
IS
and equipments were being offered from the time that the field had
begun". (Tr. 7604.) When IBM announced its unbundling decision in
1969, Honeywell conducted a study to decide what action to take.
The study recommended that' Honeywell not follow IBM. Spangle
testified that there were several reasons for this:
(1) Honeywell was not set up administratively to handle
the charge for all these separate items and to enforce their
collection throughout the field;
{2h it was not-certain of . the "contractual arrangement"
it had with its customers, and was concerned that some cus-
tomers would feel that it had contracted to furnish the items
which might otherwise be unbundled;
(3) it also "hoped to gain some temporary market advantage
. . . because we thought there would be quite a bit of resistance
to this change by the customers and prospects, and that because
of that we might be able to get some customers that we otherwise
would not have been able to get". (Spangle, Tr. 5086-89.)
Instead of unbundling, Honeywell increased its prices
19 1 slightly, since it believed that IBM's change would be regarded as a
20 : ! price increase. (Spangle, Tr. 5089.) Honeywell then began to
2x ;i advertise its "package pricing" as its "same old bundle of joy . . .
22 1 once in a while you move ahead just by standing still". (DX 13713.)
23 \ &" Product and Service Acquisitions and Expansion.
\
?4 j Prior to 1966 Honeywell had developed a series of small, high-speed
25 ' general purpose digital computers to enhance its capability to
-632-
T-3 i
provide control systems integrated with instrumentation of its own
manufacture, a related business which Honeywell had been involved in
for many years. (DX 13849/ P» 20.) In 1966 Honeywell acquired the
Computer Control Company , which at that time was a leading manufacturer
of such small, high-performance hardware,* and established it as
Honeywell's Computer Control Division. (DX 199, p. 3.) The Computer
Control Company products included the DDP-116, DDP-416 and DDP-516
computer systems and line of memories. (DX 199, p. 32.) The
computer systems were used by customers at the time in communi-
cations switching, engineering and scientific applications. They
were also offered to OEMs, who built systems for typesetting,
plotting and freight yard distribution applications. ( Id. )
Honeywell applied the "advanced digital techniques"
, gathered from the acquisition of Computer Control to its own products
in the industrial process control area (see, e.g., DX 200, p. 24) as
well as to other areas of data processing. For example, the DDP-516
was offered for time sharing, communications and medical applications..
(DX 7561, pp. 5, 9, 10.) The 516 was made available for use aboard
ships, aircraft and vans. The modified DDP-516s had all the capa-
bilities of the standard commercial version — software, price, delivery,
1
2
3
4-
3
« i
7
a
ia
ii
12
13
15
16
17
IS
IS ;,
20.
•i
i
21-1 •
j * See, e.g., DX 4917, a 1965 letter from John P. Abbadessa, Con-
22 1 troller of the U.S. Atomic Energy Commission, to a number of computer
j manufacturers requesting that the AEC's Oak Ridge National Laboratory
23 • be accorded the benefits of the manufacturers ' educational allowance
{policy. The recipients were CDC, GE , RCA, Honeywell, IBM, Burrougns ,
2d ! j Sperry Rand, SDS , Philco, Bunker Ramo and Computer Control Company.
:l
25 I
I
i
1 -633-
flexibility and proven design — while meeting the operational re-
quirements of military, marine and other users. (DX 7727.) Rug-
gedized DDP-516-s were used by the Coast Guard to gather data for
weather forecasting. (DX 756 L, p. 10; DX 9087.)* American Airlines
used 516s to control IBM 1977 terminals within its passenger services
system* (O'Neill,. Tr. 76000.) The 116 was sold to Bunker Ramo for
the "control and filing of up-to-the-minute freight booking informa-
tion" on. airline passenger planes. (DX 5789.) The 116 was also
used as part of a railroad car classification system by Westinghouse
Air Brake Company and in process control applications by the Brown
and Williamson Tobacco Corp. ( Id. , p. 7.)
In 1969 Honeywell introduced its first expansion of the
old DDP line: the Honeywell 316. Honeywell called the 316 a
" minicomputer "** and a "general purpose digital computer". (DX 123,
pp. 1-2.) The 316 had a full line of peripherals and was offered
for real-time control, data acquisition and communications applications ,
and as a front end for commercial computers made by others. (DX 7583.}
Honeywell later incorporated these smaller computers into
larger computer systems offered by it and offered them for business
\
}
)
L ;
T j
»
1
!
!
6 i
A
si
I * The Honeywell press release stated that "using a general purpose
- j machine rather than specially designed systems formerly employed
i will let the Coast Guard apply computers to many of its activities
2 \ at sea". (DX 9087.) We recognize that this press release is not in
., evidence, but we rely on it both because it represents a contempora-
2 ■ neous statement by Honeywell describing its products and because the
facts involved are independently corroborated by DX 7561.
! ** Spangle testified that a "minicomputer" is a "small general
15 i purpose computer, and small is a relative term, smaller than other
1 computers." (Tr. 4916.)
i
i
:i -634-
I
2 J
f
5
7
3
a
10'
11
data processing. It also sold these same smaller computers to its
Control Systems Division for incorporation into systems which they
in turn resold, and to outside buyers for use in specialized systems.
(Binger, Tr., 4540-42 > 4689*91; Spangle, Tr. 4916-18, 4930.)
Early in 1968 Honeywell combined its EDP Division and
Computer Control Division into the Computer and Communications
Group, to "bring into one organization those related activities that
are essential to the computer and computer-oriented business". The
new group was given the mission to involve Honeywell in the "total
information systems market". (DX 201, p. 31.)*
In the late sixties Honeywell expanded its EDP services
J2J. offerings too. It organized an Information Services Division,
offering "a broad range of integrated remote access computing services
; and contract software, as well as general data processing and
continued improved services for Honeywell EDP customers". Sixteen
data centers were opened around the country. (DX 123, p. 33.) The
centers used a Honeywell 1648 for time sharing. This system was
comprised of several Series 16 computers and was introduced by the
,, competed with the IBM 360/25 and 360/30, the DEC PDP 10 and TSS/8
1 (based on the PDP 8) , and the Hewlett-Packard 2000 A and B. (Binger,
21 ;| -
i Tr. 4593-94.)
22 i
r \
" 1
,1 * This Group was Honeywell's contribution to Honeywell Information
2- i Systems, the newly formed subsidiary of Honeywell which was merged
J with part of General Electric 's computer operations. (Binger,
25 ; , Tr. 4531.)
-635-
In sum, Honeywell's EDP operations grew steadily throughout
the 1960s* Honeywell entered the EDP business overseas in 1964
(DX 132/ p. 11) and its international EDP operations more than
doubled in 1965. Manufacturing of the 200 series in Scotland also
began in 1965. (DX I3849> pp. 28-29'.) Its Series 16 was manufactured
in the Scottish plant and in Japan by its licensee the Nippon Electric
Ti Company. (DX 201, p. 36.) Honeywell had been rapidly building its
8 ;| international computer sales force (DX 132, p. 13) ; by 1965 marketing
Q
Z
2
.4
5
.6
and service organizations existed in Australia, Canada and Western
Europe (DX 201, p. 36) and by 1969 Honeywell people served 95 percent
of the "world's computer market" . (DX 132, p. 13.)
By the end of 1969 Honeywell reported that its Computer
and Communications Group continued to be the fastest growing area of
its business; in fact, its computer and communications business was
growing faster than the industry, and its rate of profitability
increase was exceeding its growth rate. (DX 123, pp. 7, 28.)
,- | Revenues for the first nine months of 1969 increased 33.5 percent
over the comparable period in 1968. (DX 132, p. 11.) Binger stated
in a speech given in 1969 that domestic computer operations had been
*
13 :i
,_ .{ profitable for four years and overseas operations for two (id; see !
M j I
:} also DX 123, p. 7) ; indeed, the computer business was "solidly profit- ;
; ! ' I
;! able". Binger thought that the only possible factor limiting Honeywell's
— ' i !
„- ; { growth was the shortage of qualified people: ;
23 :i !
j "... we are not technology limited, we are not capital !
24 ! limited, we are not basically market limited. We may at some ;
I point be people limited to some extent." (DX 132, p. 17; see
-- .j Binger, Tr. 4818-20.) i
;| -636<
L
2
3
In 1969 the Computer and Communications Group employed 24,000 people
worldwide, compared with 18,000 a year before. Over one-half of
Honeywell's research, development and engineering dollars were spent
in the computing area. (DX 123, p. 33.) Between 1963 and 1969
Honeywell's domestic EDP revenues had increased from $27 million to
$210.8 million, a more than sevenfold increase in seven years. (DX
3
j . jj 8224, p. 387; DX 14484, p. 1? DX 8631, pp. 31, 37.)
3 !
lo ;
u
12
13
14 i
16 3
17 j
13
13
2Q I
22 :s
t
2*1
■i
•t
?* i
-637-
y
i
i
l
c i
44. Burroughs . The story of Burroughs in the period 1964
to 1970 is one of a company that turned itself around, going from
predictions . of failure to success . A slow starter in computers , by
1964 Burroughs had still not 'transformed itself into much of a com-
puter company, having developed and marketed relatively few EDP
products, (See above, pp. 227-28 •) Beginning in 1964 Burroughs
shook up its operations, reduced expenses and, while remaining profit-
able, increased its investments in research and development. The
results were a proliferation of new products , substantial growth and
increased profitability over the decade.
a. Burroughs in. 1964; Problems and Changes. Burroughs 1
situation in 1964 did not look promising for future growth in the
computer industry. Indeed, as R. W. Macdonald* wrote in 1975:
"[I]n 1964, some analysts who observed the developing computer
industry, had serious doubts about the ability of Burroughs to
survive in the new environment as a computer company. Even some
members of our own Board of Directors were concerned, and a
ff it highly respected financial journal predicted flatly that Burroughs
i! either would have to merge into another company or fail . " (DX
it 427, p. 2.)
1
j Those serious doubts were based on two factors : Burroughs ' mediocre
S '■!
;} record in computers and the perceived strength of its competition.
si
Burroughs' record in computers as of 1964 was not strong.
\ * R. W. Macdonald, a director of Burroughs since 1959, in 1964
— j became Executive Vice President, in 1966 became President and Chief
■j Operating Officer and in 1967 became President and Chief Executive
2 j Officer. (Tr. 6882-83.) Macdonald testified at trial by deposition.
:?
-638-
L
Z
3
*i
f
« !
7
3
9
ia
LI
12
13
It had begun., but had not completed, the "major transformation" from
mechanical office equipment to electronic computer technology.
(DX 427, pp. 3-4 r see above, pp. 227-28.) Its computer product
line was limited, consisting mainly of the B 200, B 5000 and D 825
computers* (See above, p. 227-) its financial record since
1961 had been poor; its revenues had "remained on a plateau" and its
earnings were "unsatisfactory". (DX 427, p. 3.)
The doubts about Burroughs ' future were also based on the
"size, profitability and technical achievements" of the "many compa-
nies [who] had aspirations to be mainframe manufacturers". Regarding
the size of those companies, Macdonald wrote:
"We faced giants such as RCA, with 1964 revenues of over
$2 billion; Honeywell, with over $600 million; Sperry Rand,
with its Univac Division, with $1.3 billion. IBM in those days
had revenues of over $3 billion, but IBM was not the largest
* . jj company we faced in terms of total revenues. General Electric,
^ ; with serious intentions and a major program in computers,
already was an industrial giant with revenues in excess of
$5 billion." (DX 427, p. 3.)
10 l| In contrast Burroughs' total annual worldwide revenues were less
- 7 |l than $400 million. (DX 10260, p. 22.)
-^ lj With regard to technical achievements, Macdonald wrote
*►- S that by 1964 IBM was "well on their way to development of a truly
impressive research and development capability" and "General
21 :! Electric had been exploring the uses of the electron for years in
22 '1 both electrical and electronic applications". By contrast:
23
"Although [Burroughs] had been engaged in electronic research
and had achieved initial success with a few very advanced
24 i new products , the products on which our revenue and profits
depended remained primarily mechanical." (DX 427, p. 3.)
-639-
1
z
3
5
a
7 |
8-!
9-
10
11
Starting in 1964 and continuing through the 195 0s
Burroughs set about to achieve its objective of "profitable growth"
and "moderate growth commensurate with maintaining profitability" in
4* j computers . (Withington, Tr, 56732; DX 10262, p. 6.) As a first
step, in 1964 Burroughs' President Ray Eppert formed the Profit
Improvement Committee. The Committee was to consider reorganization
"with respect to all aspects of marketing, manufacturing and engineer-
ing operations, and the establishment of clear product development
objectives". Its "primary charge was the swift improvement of the
company's profitability". (DX 427, p. 5.)
The changes instituted by this Committee (of which
75 I Macdonald was a member) and further changes instituted by Macdonald,
who, in 1964 was given "broad administrative responsibilities" (DX
10260, p. 4) and in 1966 became Chief Operating Officer (Tr. 6883),
were intended to accomplish two things: first, reduced expenses, and
second, improved development of computer products. (Macdonald, Tr.
If 6883-91; DX 427, pp. 7-8.)
i| (i) Reduction of Expenses. The Profit Improvement Committee
i
ij found that Burroughs ' " * problems ' lay in the efficiency of its
| operations" and not in "spending levels associated with research
i
■I
•j and development". The Committee instituted several changes* to
13 i
i
>
14*!
15
16
17
IS
19 *
20
21 ;;
22 ;i
„ ij * These changes contrast with the policies implemented at RCA
25 i during the late 1960s. At RCA expenses increased, the importance of
j financial controls was not emphasized, product development was not
2"" ij encouraged, and market share rather than profitability was considered
•! the goal. (See pp. 531-89 above.)
1
ii
■:l -640-
L
Z
3
4t
3
S
7
3
9
ia
II
12
13
14
15
IS
1?
IS
19
increase efficiency:
First , the productivity of the sales force was increased.
To do this, the Committee reduced salaries and commissions for
salesmen, reorganized the sales organization* and moved unpro~
ductive salesmen out of the division. Burroughs found that:
*The combined effect of these organizational changes
gave us the equivalent of adding 500 highly productive
salesmen— with no increase in budget costs." (DX 427,
p* 7.)**
Second , the Committee found that manufacturing costs
"had been increasing as a percentage of revenue every year for ten
years" , and the Committee undertook to reduce those costs .
It did this by reducing the number of managers at its plants,
specializing the plants by products, introducing a series
of financial controls, designating each marketing district
a "profit center", modernizing existing facilities and build-
ing 17 new plants. With these changes, by 1966 manufacturing
costs were reduced by more than five percent of revenue and
continued downward in ensuing years. (DX 427, pp. 7-8.)
Those reductions of expenses soon benefited Burroughs:
20 > * Burroughs created additional sales offices, established sales
;| zones, reduced the number of salesmen reporting to each manager, and
21 |j gave each manager a personal sales territory to cover. (DX 427, pp.
ii 6-7.)
22|!
i| ** In its 1965 Annual Report Burroughs reported that its "market-
23 | ing realignment program [had] contributed to the sales success of
:[ Burroughs' business machines and systems and improved profitability
24. of the Corporation in 1965". (DX 10261, p. 8.)
25 \
-641-
L
2
Z\
\
3
a;
7
8
9-
10
11
12!
14
15 !
16
V I
18 I
i
i
19 i
20
21
22 i
23!
24!
25 !
"The combined effect of these major changes, along with
reductions in marketing and G&A expenses and other economies
resulting from stricter overall control, produced an increase
in net earnings of over 2.00 percent in two years, from $10
million in 1964, to $31 million in 1966." (DX 427, p. 8.)
(ii) Increased Product Development. As a threshold
matter the Profit improvement Committee decided that Burroughs' lack
of profitability did not result from too much spending for research
and development. (DX 427, p. 6.) In fact, the reduction of expenses
discussed above allowed increased expenditures in research and
development :
"Removal of these excesses resulted in greater profit-
ability, which in turn made more money available for research
and development, allowing us to spend more in engineering,
leading to further cost reductions. Since 1964, we have
increased our commitment to R and D each year. ..." (DX 427,
p. 8.)
In 1964 Macdonald began to have "greater influence" in
Burroughs and his "principal activity was to utilize these resources
and developments to a much greater degree than they had in the
past" . In addition, he made sure that Burroughs would "pay a great
deal of attention to product development". (Macdonald, Tr. 6886.)
Burroughs pressed ahead with its computer developments in
two ways: First, it expanded its "product program to become more of
ja full range company". (Macdonald, Tr. 6888-89.) Second, it offered
; greater capability and increased the diversity of its computer
products. (Macdonald, Tr. 6889-90.) The addition of new products,
in turn, made more money available for research and development, as
Macdonald explained:
-642-
1
2
3
4
5
6
7
3
*
10
u!
I
12 |
-I
I4|
15!
16
17
IS
IS
20
21
22
23
24
25
"In 1964 we were operating on a research budget, an R and
D budget of approximately sixteen million dollars and as we
expanded our business we were able to afford an expanded R and
D budget and as we were able to do this we expanded the range
of products which we felt we could successfully undertake . . .
CTr. 6889.)
By 1969 Burroughs' annual spending in research and develop-
ment had doubled to $35 million. (DX 10285, p. 3.)
The changes that Burroughs began in the mid-1960s, par-
ticularly its increased research and development and improved manu-
facturing capabilities, required new investment. Macdonald described
those investments between 19^5 and 1972:
"Since 1965, Burroughs had spent some $250 million in R&D.
These funds came entirely from our own resources and were used
for the development of our commercial and trademark product
line.
"Over the same period, we have also invested just over one
billion dollars to expand the manufacturing and marketing
facilities to sell the products resulting from this R&D expen-
diture. Approximately $750 million of this represented a
marketing investment. It went for facilities, inventory,
receivables and lease funding. The remaining $250 million was
for manufacturing facilities, men, machinery and equipment. I
should also point out that this billion dollars was in addition
to the $500 million that we had already invested by the end of
1965. Of the billion dollars invested over the last seven
years, $250 million was generated through retained earnings and
I) the remaining $750 million was raised in the financial markets
through loans and equity issues." (DX 426, pp. 19-20.)
b. Computer Developments 1964 - 1969. Burroughs moved to
extend both the breadth of its product line and to increase the
capabilities offered by its computer products and by the end of 1969
had succeeded in adding many new products. This discussion traces
j the development of Burroughs ' computer products in three parts : its
500 Systems, its smaller computers and its peripheral products.
-643-
(i) The 500 Systems Family. An important factor in
Z j Burroughs ' success during the 1960s was the success of its 500
Systems family. Nine systems in that family were eventually announced:
4».| the B" 500,* B 250:0, B: 3500, ,.B 4500, B 5500, B 6500, B 7500, B 8300**
; and B '85 0O- Because of problems that Burroughs, in common with
other manufacturers, experienced with its larger machines, the B
7500, B 8300, B 8500 were either not delivered or not operational at
q J customer locations, and the B 6500 was delivered late. The B 4500 was
also never delivered. (PX 5048-D (DX 14506), Pierce, p. 62.) Still,
by 1969 Burroughs, was able to report that "this family of balanced
general purpose commercial data processors have helped the Corporation
in.!
ii
13
^ j establish an excellent position in the EDP market". (DX 10264, p. 5.)
However, while Burroughs promoted the 500 Systems as a "family", they
. . ; were not machine-language compatible as was the IBM 360, but were only
compatible through the use of higher level languages. (PX 5525-A,
p. 218; DX 10264, pp. 6, 8.)
Four months after IBM announced its System/360, Burroughs
in August 1964 announced the first member of the "500" System family,
the B 5500. (DX 13920.)/ Burroughs described the B 5500 as a
15
IS
13 !
19 I
20 \
II * Burroughs sometimes promoted the 3 500 as part of the 500
2* |j Systems (DX 10264, p. 6); it was, however, more closely related to
'•I the B 200 and B 300 and is discussed in that section.
22 !■ '
I ** Burroughs did not describe the B 8300 as a member of the 500
22 ! Systems family. However, it was closely related to the B 8500 and
is therefore discussed in this section.
24
/ Withington commented that, by announcing only one new model,
25 , Burroughs had not "attempted to answer Svstem/360 across the board".
(PX 4829, p. 22.) " |
i
i
-644- I
i
5 |
si
i
r
i
I
"modular data processing system of advanced design for both com-
mercial and scientific applications in the medium to large scale
categories". (DX 10260, p. 12.) Burroughs reported that it had "up
to three times more productivity than its predecessor, the B 5000" ►
(Id.)
B 5500s were indeed used for commercial and scientific
applications, as well as in aid of the space program. During 1964 a
B 5500 "joined the famous Atlas ground guidance computer in the
nation's space program" and was used to track the Saturn missile.
Two Burroughs' B 5500s were also used in "tabulating and projecting
l± ! national election results for the American Broadcasting Company".
Those B 5500 Systems "operated in the same manner for the election
as all Burroughs' computers do in projecting business trends, statis-
tics, competitors' activities and other information on which manage-
ments make decisions". ( Id. ) By 1965 Burroughs reported that its
orders for the B 5500 had exceeded forecast and "included many
diverse applications in national and state governments, advertising,
manufacturing, shipbuilding and research". (DX 10261, p. 9.)
During 1965 Burroughs' Defense, Space and Special Systems
Group* announced the B 8500. The B 8500 was marketed for "high
12
13
L4~
15
IS
17
IS
19
2G
21
22
23
24 'i
: "The Defense, Space and Special Systems Group produces
?5 ij and markets special data processing systems and advanced products
for the military and other government agencies. It manufactures
-645-
* Burroughs had two groups that marketed computers, its Business
Machines and Defense, Space and Special* Systems Groups. In 1968
Burroughs described the functions of its Defense, Space and Special
Systems Group:
volume, time- sharing, on-line business , scientific and government
applications" and provided for "management information processing,
including the full complement of business data processing, reporting
and. message handling as weXl as centralized or decentralized scien-
super-computer systems for government, commercial, educational
and scientific applications * The Group also produces visual
display systems , memory systems and electronic components . "
(DX 10263, p, 17.)
The relationship between the Defense , Space and Special Systems
Group and Burroughs' commercial business was close and involved,
marketing and designing the same or similar products. Burroughs
described the relationship in its 19 6 4 Annual Report r
" [tlhe Corporation's programs for various military and civilian
agencies, coupled with large investments of its own in research
and development, have yielded important technological advances
which are being utilized in Burroughs ' commercial data processing
systems and accounting machines as well as in defense and space
projects". (DX 10260, p. 19.)
Similarly, in its 1965 Annual Report, it stated that:
"[t]he Defense and Space Group was expanded in 1965 to include
the development, production and marketing of custom-built
large-scale electronic data processing systems for commercial,
industrial and special applications . These systems to a great
degree now parallel the requirements of high performance com-
puters employed in major defense and space programs where
Burroughs has had many years of successful experience." (DX
:j 10261, p. 17.)
I Products designed and marketed commercially often were later
1 J modified or further developed and marketed for military use, and
i vice versa. For example, the B 5000, the foundation for the subsequent
I j "500" product line, grew out of military work (the Burroughs D-825) .
3 (Withington, Tr. 55976-77, 58527-28.) And when the Defense, Space
1 '\ and Special Systems Group was awarded a contract to produce a mobile
i| communications system for use by the U.S. Army, it modified four B
3 : i 3500 computer systems which had been developed by the Business
[Machines Group. (DX 10716, p. 8; DX 13665, p. 19.)
4, 'j
■! At the end of 1968 the Defense, Space and Special Systems Group
5 j took over the responsibility of marketing, in addition to specially
I designed equipment, all of Burroughs computer products to the Federal
'government. (DX 13665, p. 19.)
i
;i
i -646-
L
2
3
4
5 } ; monolithic integrated circuits . (DX 10261, p. 17.)
S
7
Id
II !
12 i
13
14
IS
15
tific and engineering computations". According to Burroughs, the
B 8500 was a "logical extension" of the 'concepts of "modularity,
multiprocessing and automatic scheduling programs used with the
B 5500 and D 800 series systems..* The B 8500 also made use of
By 1967 Burroughs reported that:
"Broadening customer interest in the giant self -regulating B 8500
system confirms the importance Burroughs has given the develop-
ment and production of this supercomputer. It has the unique
ability to multiprocess a number of batches of accounting
routines, solve engineering and scientific problems, and deal
with transactions as they occur, all at the same time. The
interest of potential users in the B 8500 has greatly increased
for on-line, real— time business and scientific applications."
(DX 10263, p. 9.)
However, Burroughs experienced problems developing the
B 8500, and none was ever delivered. (Perils, Tr. 2001-02; PX 504&-D
(DX 14506), Pierce, p. 62.)
It was not until 1966 that Burroughs began to turn its 500
Systems into a family of computer systems somewhat comparable in
is ;
19
20
t7 I breadth to IBM's System/360. In that year Burroughs introduced
three new "members of the 500 systems", the B 6500, the 3 2500 and
B 3500. (DX 10262, p. 8.)
Burroughs reported that the B 6500 was "taking [its]
21 ij place" between the "medium- sized" B 5500 and the "giant" B 8500
22
-„ i| * The D 800 series included the D 825, a computer developed for
" ; the military. (See above, p. 227.) Macdonald testified that
24
25
a good deal of the B 8500s "architectural concept came from the 825"
and it "was intended to be an enlargement and in terms of size and
speed from the generation of equipment which was the D825". (Tr.
7556-57.)
-647-
3*
Si
7
a
9-
10
11 :
12
13
U!
15
17 !
18 J
19 i
20 I
L I The B 6500 central processors employed monolithic integrated circuitry
Z j throughout; had core or thin— film main memories; were "equipped for
Z-\ true multiprocessing, parallel processing, and real-time and time-
4- 1 sharing operations" ,. and had a "comprehensive , automatic operating
system for program control, completely coordinated with the hardware
elements". (DX 10262, p.. 8.)
The B 6500 was not delivered until 1969. (DX 10264, p. 8.)
Even, then its "full development" was delayed by problems in its
system software. (DX 3269, p. 3.) Burroughs reported that it had
corrected those problems in 1971. ( Id. )
The B 2500 and B 3500 were released for sale in April
1966. (DX 10262, p. 3.) Demonstrations for these systems "were
made on a broad range of business applications programmed in COBOL,
including remote processing and multiprocessing under the automatic
control of the Master Control Program" . ( Id. ) Burroughs reported
these systems would be sold "in the medium-priced range". ( Id. ,
p. 8.) By 1967 Burroughs reported that it had received "an impressive
number of orders" for the B 2500 and B 3500 from users in "such
diverse fields as finance, manufacturing, government, retailing,
insurance and publishing". (DX 10263, p. 9.)
-, ;| In 1967 Burroughs announced the B 7500. Burroughs reported
;! that its release "stimulated interest in other EDP products and
; strengthened the Company's position in this highly competitive
field". (DX 10263, p. 11.) However, the B 7500 was never delivered.
(PX 5048-D (DX 14506), Pierce, p. 62.)
24
25
-648-
L
Z
Z
In its 1968 Annual Report Burroughs reported that during
1968 it had installed the B 8300, "part of the B 8500 development
.program;" , to provide "a central passenger reservation system for a
4^| major world airline". That installation used "three central pro-
£ !t cessors functioning under the automatic control of a single software
5 '| operating system"; there were more than 2700 input and display
j 1} terminals throughout the United States with keyboard input and
8
3
IC l
II
121
13
14.
13
10
cathode ray tubes "to display data transmitted to and received from
the computer". (DX 13665, pp.. 3, 5.)
The airline at which Burroughs installed its B 8300s was
Trans World Airways (TWA). (O'Neill, Tr. 76014.) The B 8300s at
TWA were never operational, however, because the B 8300s "could not
• accommodate the projected workload" and Burroughs "had not demonstrated
adequate availability or reliability of the system". (O'Neill, Tr.
76015.) The effort was terminated, and in late 1970 TWA sued
Burroughs for non-delivery of the B 8300. (O'Neill, Tr. 76015.)*
In 1971 TWA installed one IBM 360/75 and two IBM 360/65s to
17 .,
perform the reservations function.** (O'Neill, Tr. 76013-14.)
19 :
';' * The litigation was settled in October 1972 with Burroughs
20 ; ! agreeing to assume certain payments to a leasing company and either
ij to make equipment available or to pay a sum of money to TWA. As a
21 j result of the settlement, Burroughs' earnings were reduced by
:j $4,813,000 net of taxes. (DX 10265, p. 42.)
22 :i
'ij ** Burroughs was not the only company that had difficulty installing
23 an airline passenger reservation system during the 1960s. Sperry
! Rand was also unsuccessful at installing such a system at United
24, 1 Airlines. United, like TWA, then acquired IBM equipment. (O'Neill,
JTr. 76015-17, 76231-32.)
25]
i
'! -649-
By 1969 Burroughs reported that the production of its
"'500 Systems' reached an all-time high during the year". Burroughs
described some of the reasons for the success of the 500 Systems:
"Our systems software provides self -regulated operation
which assures Burroughs customers of maximum work output through
the techniques of multiprogramming in which a number of different
programs are handled at one time. In the larger systems,
simultaneous parallel processing of programs is achieved by use
of multiple central processors. Another important advantage to
users of our medium and large systems is a modular architecture
which enables them to add processors and increase main memory
and input/output capacity in increments as needs expand.
Upward compatibility— from one '500' Systems computer to the
next largest in size — is assured through the use of higher
level programming languages. COBOL is used for business appli-
cations on the entire range of computers. For engineering and
scientific applications, FORTRAN is used on the medium-scale
B 3500 and B 5500 and the large-scale B 6500 and B 8500, and
ALGOL on the B 5500-, B 6500 and B 8500. PL 1 will be available
on the large systems and other special languages will be added
as they are required." (DX 10264, pp. 6-8.)*
These characteristics, of course, were much the same as
those IBM had earlier employed successfully in its System/360.
(ii) Smaller Computers. Burroughs marketed its smaller
;l computers in three lines: its line of B 200 successors, its E
j ' Series and the L/TC Line.
Burroughs had introduced the B 200 in 1961. (PX 552 5A,
p. 53; see above, p. 227.) In early 1965 Burroughs introduced
a :
s
2 :i
3
'!
* Burroughs reported that it had been able to obtain a design
Sj advantage with the Burroughs 500 systems:
"[w]ith the Burroughs 500 Systems, the corporation gained an
advantage by developing the software and hardware in parallel.
Engineers in these two areas combined their efforts as the
systems were developed, closing the time lag between installation
and complete usefulness of the system to the customer. This
advantage also insures the user maximum performance of the
complete system." (DX 10262, p. 10.)
-650-
L ;
2;
i
3 !
4 i one B 300 processor with a single disk file system". (DX 10260,
5!
i
!
6 :
s
9
10
11
12
13
14- ;t
15
is ;
its B 300 data processing system which was compatible with 3 200.
Burroughs reported that the B 300 included on-line capacity and that
its "modular design provides for the simultaneous use of more than
p. 12.) By 1966 Burroughs reported that the B 200 and B 300 computers
had been:
j i " [1] eased or purchased by customers in many fields including
transportation, data processing services, photo supplies,
utilities, insurance, publishing, brewing, school systems,
manufacturing, baking, textile milling, property management,
retailing, wholesaling, distributing, government and public
service, research and finance." (DX 10262, p. 12.)
In 1965 Burroughs introduced the B 340 bank data processing
system which was smaller than, although compatible with, the B 300
system. (DX 10263, p. 11.) During 1968 Burroughs introduced its
3 500 computer. The B 500 had an automatic operating system and used
COBOL. While promoted by Burroughs as a "member of the '500' Systems
EDP family", the 500 was compatible in assembly language with the B
100, B 200 and B 300 systems but was compatible only through the use
*7 i| of COBOL with the 500 Systems. (DX 13665, p. 5.)
1 a I
*° ;i During 1964 Burroughs brought out its E 2100 computer.
19 1
^ !j (DX 10260, p. 8.) Between 1964 and 1970 it added to the E Series
20 j with the E 3000, E 5000, E 6000 and E 8000. (DX 10263, p. 11;
2- ! j DX 10264, p. 18.) The E Series were small solid state computers
jl
22 j with electronic logic and data storage. • (DX 10260, p. 9; DX 10264,
22 -j p. 18.) On the larger E Series computers, the E 6000 and E 8000,
; t
24 :j COBOL was available. (DX 10264, p. 18.)
In 1968 Burroughs took a major step forward with the
-651-
25
L
2
3
4
5
5
7
3
9-
10
11
12 !
13
14
15
IS
17
IS
IS
20
21
22
23
24
announcement of its TC 500 terminal computer. (DX 13665, pp. 1, 7.)
The TC 500 was characterized by John Jones of Southern Railway as
the first "intelligent terminal", that is, the "first programmable
terminal . . . that had in it a processor, a general purpose pro-
cessor with memory and input and output, that could be programmed to
perform in some way as the user desired as opposed to being hard
wired". (Tr. 79044-45.) The TC 500:
" [h] ad a keyboard for an operator to input data and a printer
on which data could be printed, a character printer, and a
processor inside of it which could be programmed to give that
device any particular characteristics in its. operation, as well
as do other processing of the data as it was entered or before
it was printed." (Tr. 79044.)
Burroughs similarly described the TC 500 in its Annual Report:.
"An internally programmed computer using integrated circuitry
and dis3c memory, the TC 500 can operate as a data communications
terminal on-line to a central computer, or function off-line
independently. ... In addition to data communications,
they can edit and format information and perform functions
which previously had to be handled by the central computer."
(DX 13665, p. 7.)
In 1969 Burroughs introduced its TC 700 and TC 310 terminal
I
} computers. The "TC 500 and TC 700 have their own computing and
j memory capabilities. They also edit and format information for most
| economical transmission to a central system. The TC 310, in multiples,
j is connected to a data controller which then performs the formatting
; and other necessary operations prior to data transmission." (DX
'' 10264, p. 14.)
Also resulting from the same engineering as the TC computers
was the L 2000 computer. Introduced in 1969, the L 2000 was a
\
^ ; computer, designed for billing, as to which "the addition of a data
-652-
L
2
3
5
S
7
3
10
11 j
i
13 i
i
14- I
i
i
is ;
is I
17 !
13 ■
communications unit converts it to a terminal computer able to
communicate with a central computer system". (DX 10264, p. 18.)
COBOL was available for the L 2000. ( Id. )
Macdonald described the L/TC series as follows:
"These internally programmed machines are programmed in COBOL
and can operate under operator control or under program control.
ti
• . • .
"These small systems are, in terms of what they can perform,
small full-scale computers." (DX 10285, p. 6.)
(iii) Peripherals. During the years from 1964 through
1969 Burroughs improved upon its existing peripheral equipment. It
introduced several models of improved card readers, printers, sorter-
readers, tape transports, multi-tape listers, and tape drives. (DX
10261, p. 11; DX 10264, p. 10.)
By 1964 Burroughs had developed and was marketing a disk
file with a head-per-track. (Withington, Tr. 56244; DX 10260, p. 10.)
This head-per-track file had a slightly faster access time and a
slightly higher cost per unit of storage than the movable head
devices. (Withington, Tr. 56244-45.) During the mid-1960s Burroughs
'! found that its disk drive was "a significant factor in the growth of
19
2Q 1 the Company's business in EDP systems". (DX 13665, p. 5; see also
-, ":| PX 4834, p. 31.)
22 l! However, in 1962 when IBM introduced its 1311 disk drive
si
-- ! | with a removable disk pack, Burroughs did not offer a disk drive
- A ;[ with a similar removable pack, nor did Burroughs offer such a disk
„ g j drive after IBM followed this announcement with the introduction of
-653-
i
t!
3
3
a
z
■**
-5
1 «r
.3
L7
LS;j
the 2311 and 2314 disk drives. Where Burroughs' customers wanted
the advantages of a removable disk pack, Burroughs sought to convince
them to keep their files on magnetic tape and to load and unload the
files on to the Burroughs' fixed pack drives. (Withington, Tr.
58802.) Finally, in the late 1960s, Burroughs arranged to acquire
5 ;' disk drives with removable disk packs from Century Data, and in 1970
7 : ;j it began marketing those disk drives as part of its computer systems .
(PX 4445, pp. 7-8; DX 10716, p. 12.)
Burroughs introduced new peripheral equipment during the
1960s. In 1969 Burroughs introduced a new electronic reader/sorter
which handled documents both optically and magnetically encoded and a
new computer-output-to-microfilm system. It also announced three
new encoding devices : the Series N keyboard-to-magnetic tape data
encoding machine; the A 149 peripheral card punch, the A 150 keypunch,
and the A 160 verifier for punched card encoding; and the Series S
"general purpose character encoding machines . . . designed to
encode unit documents ... to facilitate electronic reading by
high-speed recognition equipment". (DX 10264, pp. 10-14.)
c. Burroughs at the End of the 1960s. By the end of the
I decade the changes Burroughs had instituted in 1964 had begun to j
ZO "! |
i achieve Burroughs' objective of "profitable growth". (DX 427, p. 8.) j
21 1 I
:j Burroughs had reduced costs and increased efficiency in its manuf ac- j
^ 1 !
:l turing and marketing operations (DX 427, pp. 6-8), and it had increased j
23 ■
jits expenditures in research and development. (Macdonald, Tr. 6 889; j
2- i I
! DX 427, p. 8; DX 10264, pp. 4, 6.) Aided by those changes, Burroughs j
1* \
-654-
L
2
3
3
6
7
3
9
ia
ii
12
13
14;
15
16 II
had expanded its product line in terms of both range and the capa-
bilities offered. From a few mid-size computers in 1964, Burroughs
delivered several complete lines of computer systems ranging from
small (E Series; L/TC Series) to very large (B 6500) by 1969. Its
numerous new product offerings were reflected in its 1969 Annual
Report which described its "broad line of products for the data
recording, computing and processing market" including:
"[c]omputer systems, memory sub-systems, peripheral
input and output equipment, data encoding equipment, data
communications terminals, accounting systems, calculators and
adding machines, business forms and office supplies, custom-
designed electronic systems, and data display devices. This
extensive range of products represents one of Burroughs basic
strengths for continued growth in the rapidly expanding data
processing industry." (DX 10264, p. 2.)
And Burroughs was also continuing its technological development in
intelligent terminals, an area that would become very important in
the 1970s. (See DX 10264, p. 14.)
It was clear that Burroughs ' management understood the close
interrelationship of its extensive product line. In a 1969 presenta-
^' ij tion to the New York Society of Security Analysts, Ray Macdonald
stated, concerning the relationship among various computer products,
13
19 ;j
,j that:
^ J .| "In 1967, I said that when I had the next opportunity of •
•j addressing this group we might refer to electronic accounting
2- !! machines, electronic accounting systems, terminal units and
•I electronic computers as one continuous market from small machine
2~ jj to giant computer. This blending of several markets into a
__ i( single broad market has now become more evident." (DX 10285,
^ :| P. 5.)
** ' Burroughs' financial results, in turn, reflected the
25 ; proliferation of its computer products. From 1964 to 1969 Burroughs'
-655-
total corporate revenues did not quite double, increasing from $392
million to $759 million. During the same time its domestic EDP
revenues increased from $61 million to $260 million and its corporate
»
profits jumped 500 percent. (DX 8224, p. 1; DX 10260, p. 5; DX 10264,
5 I P. 5.)
Writing in 1975, Macdonald looked back on the results of
7 | the changes that Burroughs had instituted in 1964:
3
ixi
i
12.1
13 !
i
14.!
15
16 ;!
17 j
IS !
13 I
20 ;
21 ||
22 -j
23 j
24 :|
"Our revenue has doubled every five years, and today, at
$1.5 billion, is four times its level of ten years ago.
"Our net earnings have increased by 14 times during
the 10-year period, and this is the best record of growth in
the mainframe computer industry.
"Our manpower worldwide has increased from about 34,000
to more than 51,500. We are operating 54 plants in ten countries
and two more plants are under construction." (DX 427, pp. 2-3.)
By the beginning of 1970 Burroughs had made up much of the
ground it had lost during the 1950s and early 1960s, and was well
situated for even greater success in the 1970s.
-656-
1 45. National Cash Register. Historically, National Cash
2 [Register (NCR) was a company that had concentrated on marketing its
i
3 ! "traditional products" — cash registers, accounting machines and adding
4 | machines — to customers engaged in retailing and banking. (DX 344, p.
5 ;1; DX 372, p. 1; see pp. 229-31, 236 above.) By the beginning of
6 11964, while continuing to concentrate on customers in those areas,
7 NCR had introduced and was marketing two models of its second
8 generation 351 computer system which had been announced in the early
9 1960s. (DX 344, p. 14; DX 382, pp. 3, 10.) At the same time NCR
10 ! was actively expanding "the functions of its traditional products".
i
11 (See above p. 241; DX 344, p. 1.) In 1964 NCR's domestic EDP revenues
12 ($46.3 million) accounted for only about 13 percent of its total
i
13 domestic revenues. (DX 361, p. 22; DX 8224, p. 3.)
j
14 | The story of NCR in the years between 1964 and 1970 is that
i
15 of a company wishing to maintain its traditional business and only
16 I gradually adding the increased capability offered by computers. This
17 i| desire gradually to develop computers to support its traditional
ii
ii
IS j business was expressed by the President of NCR, Robert S. Oelman,*
19 j in a November 1964 speech. He stated that the company had "recently"
20 undergone "the most significant change in [its] long history . . . the
21 advent of electronic data processing." (DX 342, pp. 2-3.) However,
22
23 j * Oelman, along with J. J. Hangen, were the two witnesses called by
j plaintiff from NCR. From 1964 until he left NCR in 1973, Oelman was
24 j Chairman and Chief Executive Officer of NCR (Tr. 6117) , and from 1964
! until 1972 Hangen was Vice President of Finance. (Tr. 6233.) At
25 i the time he testified Hangen was Senior Vice President of Corporate
Affairs for NCR. (Tr. 6239.)
-657-
1 !
I
i
2 !
3J
i
5 I
i
s !
7 |
3 !
i
9 !
i
111
12 |
13j
i
14.!
.1
i
15 i
t
I
16 :
17 !i
Oelman explained that this change did not mean the demise of NCR's
"traditional products" — cash registers, accounting machines and adding
machines — for two reasons: First, the traditional products were
"being integrated" into electronic data processing systems,- the tradi-
tional products serve as an "input medium" for data and are tied into
"the mainstream of the data processing revolution". Second, NCR could
use "new technologies to add important machine features and to improve
overall performance" of its traditional products. ( Id. , pp. 3-6.)
Thus, NCR, rather than recognizing that computers were going
to obsolete its "traditional business" (as IBM had in the 1950s) and
committing itself to the new technology, chose to split its resources
between computers and its traditional cash register and accounting
machine products. (See, e.g., DX 361, p. 1; DX 370, p. 16.) In its
1966 Annual Report NCR reported that:
"The Company's R&D program is designed to achieve two
basic objectives:
"1. To improve NCR's traditional position of leadership in the
control register, accounting machine and adding machine
markets ;
^ it "2. To gain for the Company an increasing share of the rapidly
:| growing market for computer systems and related equipment."
19 ; (DX 370, p. 16.)
2$ [j Outside observers also reported on NCR's desire to proceed
.•I
2- || gradually in computers. Withington described NCR as following a plan
22 !i during the 1960s to proceed "methodically" in computers by using
^ I computers "to complement its existing product and marketing positions".
^ NCR did this because
25 !
1
;l -658-
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
IS
19
20
"[t]he risks and investments involved in introducing highly
innovative products to rapidly achieve a major share of computer
shipments do not appeal to NCR, and that as long as the company's
overall position, growth, and profit objectives are supported the
company's computer market share is not a primary objective." (PX
4834, p. 34.)
Similarly, an analysis of NCR. by IBM employees in IBM's Market
Evaluation Department observed that in 1967 NCR was "still in the
process of establishing itself in computers [and its] management is
not prone to risk ventures". (PX 2050, p. 4.)
Developments at NCR during the period from 1964 to 1970 show
a company improving its products gradually and trying to avoid taking
the risks of producing innovative products. From 1964 to 1968 NCR
introduced only a few improvements to its second generation equipment.
During the summer of 1964, NCR announced a follow-on member of the 315
family, the 315 Rod Memory Computer (RMC) . (DX 361, p. 14; DX 401, p.
2.) The 315 RMC used thin-film memory technology and was compatible
with other computers in the 315 line. (Hangen, Tr. 6 314; DX 361, p.
14.) Multiprogramming for the 315 RMC was announced during 1966. (DX
370, p. 16.) During 1965 NCR announced the Series 500 computer, a
general purpose computer which attempted to combine "magnetic ledger
bookkeeping with various combinations of punched card, punched paper
tape or optical equipment". (Hangen, Tr. 10 402; DX 361, p. 13.)
21 li
;! NCR also continued to make changes to its existing products .
|j Despite the fact that the CRAM file had' been superseded by the disk
ij drive (discussed above at p. 23 5; see Withington, Tr. 56469-70,
24
25
■ !
56511) , in 1966 NCR announced a more powerful version of that product
(DX 370, p. 16) rather than replace it entirely with disk drives.
-659-
1 i
i
2!
i
1
3 i
1
1
4-1
1
5 !
1
!
i
1
3 i
9 i
i
10 i
i
12 J
13 !
14 !
i
15 !
Following the trend started by System/360, NCR reported in 1966
that it was increasing the modularity offered by its 315 computer
family:
"When NCR announced the 315 computer family, four basic
processors, three memory sizes, and 12 peripherals were offered.
Today, 315 users have a choice of nine different processors,
eight memory combinations and some 60 peripheral units. Expan-
sion of this flexible computer series in 1965 included three new
magnetic tape units , three new high-speed printers and a new
communications controller. This latter device permits up to 100
communication lines serving input or outDut devices to be linked
directly to a 315 ► It greatly" extends the power of the system to
receive inquiries from remote locations and transmit answers . "
(DX 368, p. 6.)
It was not until March 5, 1968, almost four years after
IBM's announcement of System/360, that NCR introduced its third
generation computers , the Century Series . The first models announced
were the Century 100 and 200, and NCR stated that it intended soon to
announce a Century 400 which would be capable of performing time
sharing. (DX 348, p. 1.)
,c || The Century 200 was "designed for batch, real-time, and
17
scientific processing". (DX 469, p. 2.) NCR offered the Century 100
1fl II and 200 Systems, in addition to sale, on one, three or five year
!j rental terms.* (DX 348, pp. 1-2.) Each system was marketed with
jj
'\ minimum amount of main memory, a card reader or paper tape reader,
2o '.!
-. jj printer and disk drive. (DX 348, p. 1.) Other available peripherals
. Ji
;! included CRAM, a MICR sorter-reader, an optical journal reader,
ii
-- Ii punched card units, and visual display units. (DX 421, p. 17; DX
24 -j
1 * By April 30, 1969, over half of the orders for the Century Series
" |j were for a five-year term. (DX 372, p. 4.)
I
;| -660-
1 ;
i
i
2 !
3
3
6 I
7
3
9-
10
II !
12
13
14-
15
la
17
469, p. 14.)
NCR promoted the Century Series as its "most important
new line of products" (DX 366 , p. A), asserting that it incorpo-
rated many advances over its previous machines, including:
(1) The Century Series continued the use of thin- film
main memory introduced on the 315 RMC. NCR called this an
"important 'first 1 ", making the performance of the thin-film
memory available at a lower cost* (DX 366, p. 5.) Within
about a year after the announcement of the Century Series,
NCR replaced the thin-film memory with core memory. (Hangen,
Tr. 6329-30.)
(2) The Century Series used integrated circuits "throughout
all Century computers and peripherals". (DX 366, p. 6.)
(3) The Century Series provided for "complete upward
compatability" so that "as a user's needs increase, more
powerful processors can replace original units as required".
(Id., p. 8.) '
|t (4) The Century Series included more advanced peri-
j) pherals — including, for the first time, disk drives: "The
|| philosophy" of the Century Series "is that the disc concept
;| is an integral part of all members of the family". The
jj Series also included a new high-speed printer and, yet again,
an improved CRAM unit. ( Id. , pp. 6-7.)
(5) The Century Series had the capability to use both
23
24 ;
-661-
I!
!
3!
i
♦■I
I
I
s j
7j
i
3!
»l
io!
12
13
14
15
16
17
IS
19 it
20
COBOL and FORTRAN* programs among others.**
(6) The Century Series provided for standardization in
design including standard cabinet frames and panels, power
supplies and cable connections. It also provided for standard
interfaces so that "the many peripheral units available with
Century processors can 'interface' simply, and in a wide
variety of configurations" . ( Id . ,. p. 7.)
Of course, while these features represented improvements over
NCR's prior products, all these features, with the exception of the
soon to be discontinued thin-film memory, had been included in IBM's
System/360 four years earlier./
In its Annual Report for 1968, NCR announced its marketing
plans for the Century Series:
"Over the years NCR has established itself as a
leading supplier of business systems to thousands of
manufacturing concerns, construction companies, whole-
salers, schools, hospitals, utilities, hotels and
motels, business service firms, and local, state and
federal government offices.
* According to Hangen, FORTRAN would be used by NCR users to perform
scientific applications. (Tr. 10604.)
** On software development, NCR reported:
!l
"Basic computer operating software as well as standard
Zi |! application programs have been prepared concurrently with
equipment development. This has insured full program com-
patibility, plus a proper balance between 'hardware' and
'software' capabilities." (DX 366, p. 10.)
22
23
24
25
/ During this period, perhaps because of NCR's late response to
System/360, NCR "as a general rule . . . attempt [ed] to price [its]
products slightly less than the comparable IBM system", that is, "5
to 10 percent less". (Hangen, Tr. 6350-51.)
-662-
1 1
j
i
2 >
3 !
I
5 I
6
7
3
9
10
11
12
13
14
15
IS
"The advent of the Century Series computer family-
has multiplied the company ' s opportunities in these
fields . As users of NCR accounting machines grow and
their data processing requirements increase, a Century
100 computer system can meet these greater needs just
as the Century 200 can serve the larger organization.
At the same time however, with thousands of new small
businesses being established each year, the market for
accounting machines has continued to grow."
"The largest single market for computer systems is
in manufacturing^ One out of every four Century Series
computers currently on order, for example, is scheduled
for use in this area." (DX 340A, p. 8.)
Hangen, then Vice President of Finance (Tr. 6233) , emphasized
in April 1969 the opportunity Century afforded to broaden NCR's market
ing thrust:
"Although we intend to continue our close relation-
ships with the retailing and financial industries, the
Century allows us to broaden our marketing thrust. We
are offering specialized Century programs for the
Educational, Hospital, Local Government (including Law
Enforcement) , and Distribution Industries." (DX 372,
pp. 3-4.)
He added that the majority of orders received for the Century were fron
17 j| "non-banking, non-retail industries". (DX 372, p. 4.)
IS Ij The Century Series was "largely responsible" for the fact j
!j I
19 ii that in 1968 NCR's domestic orders for computers increased 98% over the
l| 1
20 ij prior years. The result of the increase was that "for the first time
21 ij domestic orders for computer equipment exceeded those for either cash
22 Ii registers or accounting machines." (DX. 340A, p. 2.) To meet the
23 '• demand NCR expanded the Electronics Division plant facilities by 50%
24 'j
s!
ij -663-
and planned a further increase in 1969. (DX 340A, ?. 2.)*
The large increase in orders had an "adverse impact" on NCR's
earnings during 1968 as the 1968 Annual Report explained:
► j- "Users of computer ' systems generally prefer to
rent rather, than buy such equipment. Thus, the intro-
duction of a major new computer family such as the
Century Series tends to have an adverse impact on
earnings initially, since the company must immediately
bear production startup, software, training and depre-
ciation expenses although revenue from rental instal-
lations is received only over a period of years . " (DX
340A, pp. 2-3.)
NCR went on to assure its- stockholders that in -future years
there would be "a highly favorable effect on earnings". (DX
340A, pp. 2-3.) Ninety percent of the Century Series systems
marketed were in fact leased. (Hangen, Tr. 6 358-59.)
Throughout the 1960s, NCR understood the importance of
support services — customer training, maintenance, systems design — in
marketing computer products. It stated in its 1964 Annual Report that:
"The user of an NCR business system buys con-
,! siderably more than the machine units which make up
that system.' In every case, an NCR systems specialist
and in many instances teams of specialists design
the most efficient system possible to meet the
customer's current and future needs, then thoroughly
train the user's staff in its use. After the system
3 i is operational, further counseling and assistance
; l including dependable maintenance are provided."
G ] (DX 361, p. 8.)
;i
"7
Similarly, it stated in its 1966 Report that:
5 i * The Century 100 was first shipped in the fall of 1968 (DX 340A,
| p. B) , and the Century 200 was first shipped in June or July 1969.
:4 i (Hangen, Tr. 6328.)
■| -664-
i
:i
'i
; !
!
1
2
3
4
5
6
7
S
9-
ia
li
12 i
13
14
15
16
17
IS
"Today, a . . . requirement for future success in
the marketplace has arisen; that is the need for
business equipment suppliers to provide additional
guidance to customers in utilization of new technologies
for operating their businesses more profitably. For in
the final analysis, the effectiveness of today's
sophisticated information systems depends upon a full
understanding of their potential at all levels of manage-
ment* To this end NCR's educational programs are being
designed not only to prepare sales representatives to
install advanced systems, but also to provide counsel
and training in management sciences." (DX 370, p. 5.)
And, with the introduction of its Century System it realized that
customers needed even more support. In its 1969 Annual Report, it
stated:
"Marketing requirements of the business equipment
industry have changed significantly in recent years .
In recognition of this , the company has taken various
steps to provide the greater degree of support which
customers need and expect." (DX 367, p. 4.)
After IBM announced its "unbundling" in June 1969, NCR's
Pricing Committee decided whether to make any changes in its pricing of
support services. (Hangen, Tr. 6364.) Recognizing that there would be
problems, "particularly [in] customer relations", the Pricing Committee
did not take any immediate action. (Hangen, Tr. 6393.) On October 1,
It 1969, NCR announced a change in its pricing structure. (DX 346.) The
^ ij announcement stated in part
20 :[
"NCR believes that each user of its computer systems
must be provided with a certain essential amount of
software, systems support, and educational services if
)! he is to successfully install the system and begin
22 !} to benefit from his investment. NCR believes that
this k as i c package of supporting services must be the
responsibility of the equipment manufacturer.
21
23 !
24
"In addition, NCR recognizes that there is considerable
variance in the level of support required by different
customers. This is a function of the capabilities of
-665-
f .;
■
i !
i
i
3
7
S
the customer's Internal EDP staff and of the scope
and complexity of the applications to be installed.
"Accordingly, it will continue to be NCR's policy to
provide,, as part of the basic hardware price, that amount
of software and support which will realistically insure
that a prudent user will be able to install and success-
fully utilize his NCR computer system.
"Software and support services requested above the
level which is included in the basic hardware price
will be separately priced." (DX 346, p. 1, emphasis
in original.)
NCR did not change basic hardware prices when it started to charge
separately for those support services. (Hangen, Tr. 6365.)
On January 1, 1970, however, NCR partially reversed
its unbundling decision and announced:
"After further evaluation, it has been decided not
to price all basic and applied software and not to
establish an allowance against which such chargeable
software would be applied. The NCR software pricing
plan will be to continue to establish pricing for
software products on a selective basis , considering
the value to the customer, uniqueness, and other
factors. This approach creates an allowance effect
since the more basic software offerings will not be
priced." (DX 386, p. 2, emphasis in original.)
NCR's computer data center business, begun in 1960, expanded
a :i
2 i during the 1960s, so that by 1968 there were 69 centers worldwide.
n \
u i Many customers of NCR's data centers used NCR cash registers, account-
I j ing machines or adding machines to produce "punched paper tape or
I
2. i machine readable 'optical' figures as a by-product of normal opera-
13 : tion." The customers then sent the output media to NCR's data center
i
t
IX I for processing. (DX 340A, p. 10.) The data centers were "NCR's most
t
.i
i5 j successful effort in the data processing business" in the 1960s,
i
t
i
J -666-
1
2
3
5
6 i
7
a !
i
9 I
„i
12}
13 I
I
i
I
15 !
16
according to Withington (PX 4832, p. 22) , and, in addition, proved to
be a "powerful stimulus to the sale and rental of data capturing"
devices. (DX 340A, p. 3.)
NCR's use of its traditional products as input devices for
its data center computers was an example of NCR's attempt to integrate
its traditional products with its computer systems. In 1963 NCR
reported that those products could be used with computers in several
ways:
"Many different types of cash registers , accounting
machines , adding machines and other peripheral units
are available as basic input devices for [computer]
systems . Some of these machines communicate with
computers by means of punched tape or punched cards .
Others record transactions or other data in slightly
stylized print which can be read by optical or magnetic
scanning machines * Still others can be cabled directly
'on-line' to NCR electronic data processing systems."
(DX 344, p. 3.)
NCR did very little, however, in terms of developing and
marketing on-line systems during the 1960s. During May 1969 H. M.
'Keller, NCR's Manager of Terminal Communications Products, wrote that
! in terminal and communication products NCR did "not have a great choice
ii
IS ; ' i
i to offer our prospects", and he listed only one on-line device, the 42-
11 500, a bank tellers' console. (DX 719, p. i, see Oelman, Tr. 6164.)
20 : : .
;! Keller noted, however, that a change had recently occurred in NCR's
21 ! i
! ; commitment to on-line devices:
*^ II "Before we knew that our Company committed itself
-« jl to creating and offering terminal devices for many,
lj many purposes , we may have had reasons for not encourag-
j( ing sales of on-line systems. Now that we know that
2 ' ;j NCR is committed, each of us must help to penetrate the
on-line field." (DX 719, p. 1.)
25
-667-
L j In support of that commitment NCR was investing "tremendous
i
2,1 sums of money in developing" terminal and communications devices.
3 | Keller predicted that:
} '
4. ! "To quite some extent, our future success in the
terminal field will depend upon our success with computer
5 j. sales and installations » On the other hand, the avail-
ability of a complete range of terminals will certainly
gi further enhance our CENTURY sales." (DX 719, p. 1.)
7 I NCR made a similar prediction concerning terminals in its
1
a I 1969 Annual Report:
10
11
n j "More and more people will be brought into direct
communications with computers through a variety of data
terminals and data display devices. In fact, it is
anticipated that by 1975 users of data processing
systems will be investing as much or more in data
terminals and related communications equipment as in
--j the central computer itself. This will create major
! new opportunities for the business equipment industry
1- I and particularly for companies such as NCR which has
j extensive experience in data entry devices." (DX 367,
l4 J PP. 9-10.)
2* i And:
IS ;
i
17 J
13 I
19 I
20 :
21 I
22
24 1
25
"A decade ago, almost all business machines were
sold as free-standing equipment. Today, many of these
products as well as entirely new types of equipment are
linked together as "total" systems to meet individual
customer needs. Such systems often include arrays of
compatible computer equipment including communications
networks." (DX 367, p. 19.)
Those predictions turned out to be accurate. During the
1970s, NCR found that "the capabilities and price/performance of its
jl terminals [were] an important factor in convincing users to take NCR
computer systems." (Oelman, Tr. 6183; see also below pp. 998-99.)
By 1970 it was plain that NCR had oroceeded "methodically" in;
j
the computer business, avoiding risks but avoiding also the great j
-668-
L
2 \
3
4.
f
r
3
9-
10
Li
12
13
14- :
16 |
17 I
IS
19
success that comes with successful risk-taking. Between 1964 and 1970 ,
NCR's most significant development was the introduction of two models
(the Century 100 and 200) of a system the principal features of which
had been available on IBM's System/360 delivered three years
earlier. With this gradual development, however, NCR reduced the
chances that it would be a failure like GE and RCA and found itself
positioned to turn the corner in the 1970s, which it ultimately did.
NCR's domestic EDP revenues for the year 1969 were $179,298,000, over
five times its U.S. EDP revenues in 1963. (DX 8224, p. 3.) Even with
that growth, NCR's domestic EDP revenues accounted for only 26 percent
of NCR's total domestic revenues but double what it had in 1963.
(DX 367, p. 6.)
21 ]
•i
22 >
»1
2± I
i
?c 1
-669-
46. Control Data Corporation (CDC) . The period from 1963 to
1969 was one of rapid expansion for CDC. It added to its two principal
product lines, the 3000 and 6000 series; it expanded the applications
. capabilities of its computers to include not only a scientific emphasis
e j[- ^ut also business -oriented software; for the first time, it developed,
3
T
3
9
a
l
A \
e
.5
L7
LS
manufactured and marketed a broad line of peripheral equipment, includ-
ing OEM sales to other companies and IBM plug-compatible equipment; it
expanded its overseas operations; it made a large number of acquisi-
tions, including, most importantly that of Commercial Credit Corpora-
tion, a large financial services company; and, finally, it greatly
expanded its data center business* CDC's total EDP revenues grew from
S85 million in 1963 to $570 million in 1969. (DX 298.) Its U.S. EDP
revenues grew from $83 million in 1964 to $458 million in 1969. (DX
8224, p. 5.) Its assets increased from $71 million in 1963 to $761
million in 1969. (DX 302.) To finance that expansion, at least in
part, CDC raised over $767 million between 1963 and 1969 through equity
and long-term debt financings. (DX 300.)
The discussion below focuses on five principal areas of CDC's
![ growth during the 1963-1969 period: First, CDC's major product
; announcements, including, principally, its 6000 and 3000 series, and
20"!
| their success; second, the increasing use of CDC computers to perform
21 i
j both scientific and commercial applications; third, CDC's expanding
t peripheral equipment offerings , both as attachments to CDC systems and
22 •
|as OEM products sold to other companies; fourth, the expansion of CDC's
7L i
j data centers; and, fifth, the way in which CDC's growth took place,
2SJ
j
■i -670-
3. '
4
5 !
i
■
7j
i
!
3-!
»i
H
11 1
12 I
i
13]
I
14!
I
t
15 !
i
le I
17 J
IS I
19 !
zz -
21 j
!
22j
23 |
24!
2S ;
i.e. , by vertical integration and acquisition.
a - CDC's 6000 and 3000 Series Offerings (196 3-1969) .
(i) The 6000 Series. CDC's most important product in the
1960s was undoubtedly the 6600 computer, announced in July 1962 in
5 ► connection with a contract let by the Atomic Energy Commission's
Lawrence Livermore Laboratory, and first delivered in September 1964
(seven months later than the date contracted for) . (JX 10, 1f 4.) CDC
Chairman and Chief Executive Officer William Norris* described the 6600
as a "very great risk" since "it was a trip into the unknown" and
* William Norris was one of the four founders of Control Data in
July 1957. (Tr. 5604.) In August 1957 he was elected to the Board of
Directors and to the position of President of CDC. In 1958-59 he
assumed the additional title of Chairman of the Board of CDC. At the
time of his testimony in 1975, Mr. Norris had been Chairman and Chief
Executive Officer of CDC for over seventeen years. (Tr. 5596-97; PX
355, pp. 11-29.) Prior to the formation of CDC, Mr. Norris was General
Manager of the Univac Division of Sperry Rand. (Tr. 5603.)
R. D. Schmidt joined CDC in 1962 as a salesman. At the time of
his testimony, Mr. Schmidt was a member of the Board of Directors of
CDC, Executive Vice President of the corporation and Chairman of its
Export Strategy Committee. (Tr. 27199-201.)
. W. Lacey had been employed by CDC for approximately fifteen
at the time of his testimony, and held the position of Senior
J
years at
Vice President of Corporate Plans and Development. In addition, he was
Chairman of CDC's Operations Committee and a member of the Board of
Directors of CDC's Commercial Credit Corporation subsidiary. (Tr.
6552-53.)
Gordon Brown, at the time of his testimony, was Senior Vice Presi-
dent of Marketing and Planning for CDC's Peripheral Products Company.
(Tr. 50977-78.)
H. W. Forrest testified by deposition. (DX 13526.) Forrest
worked in the Univac Division of Sperry Rand under Norris and moved
with Norris to CDC. ( Id. , pp. 42-44.) At the time he was deposed,
Forrest was Senior Vice President, Government Relations, for CDC.
( Id. , p. 4.)
-671-
1 i
i
2;
s\
4*i
J I Series computers . ( See Norr is , Tr . 5849-51.)
5
7
8
9-
1(1 I
111
12
13
14,1
i
15 !
16 (
it j
13 I
19 \
20 ;
testified that CDC was "betting the future of the company" on it.
(Norris, Tr.. 5616.) But, as did IBM with System/360, CDC received
considerable returns on its "bet". Despite early problems with the
6600, CDC ultimately was successful with it and with the other 6000
On December 15, 1964, some eight months after IBM's System/
360 announcement, CDC formally announced the "6000 Series", then con-
sisting of the compatible 6400, 6600 and 6800 computers.* (DX 319.) In
the announcement press release, Norr is described the 6000 Series as
"the industry's most extensive product line of super-scale compu-
ters . . . provide ing] business, industry, science and government users
the most comprehensive range of software and system compatibility ever
announced in the computer industry." (DX 319, p. 1.) Purchase prices
for typical 6000 Series systems were announced as ranging "from less
than $1 million to several million" with rental prices from $25,0 00 to
$150,000 or more per month.** (DX 319, p. 3.)
By the end of 1964, CDC had received "possibly five or six" -
orders for the 6600 (Norris, Tr. 5624), although top officials at IBM
had believed as early as the Fall of 1963 that as many as 10 accounts
were then planning to order CDC 6600s. (JX 10, II 9.) Deliveries of
21 ijthe 6600 were delayed, however, due to unanticipated technological
li
22 :i
^5 ;
* The 6800, however, was never delivered. (Norris, Tr. 5967; JX 10,
If 34.)
** :j *'* CDC later announced two more models of the 6000 Series, the 6500,
5 - ; i announced in March 1967, and the 6700, announced in May 1969. (PX 355,
jpp. 36-37; see Norris, Tr. 5626.) Norris described the 6500 as "actu-
ally two 6400 ! s connected together" and the 6700 as "somewhat more
I powerful", being "basically two 6600s." (Tr. 5626.)
-672-
problems in 1964, 1965 and 1966. (See JX 10, 1! 34; Norris , Tr. 5853-
54.) By 1966 CDC had solved its technological problems and reported to
its stockholders on the problems which had occurred in its development
of large systems:
" . . * The development and manufacture of very large
computers are extremely difficult, and severe technical problems
are inherent in the process. In a past stockholder report,
it was emphasized that estimating completion dates of very
large computer developments is becoming increasingly difficult.
Last year at this time we believed we had found solutions to
the major technical problems in connection with the 6600.
Experience since then has proven that, while this was generally
true, the estimate was in error on the time and effort required
to make the necessary changes in the equipment and programs.
The process took longer than anticipated; as a result,, we
incurred increased penalties for late delivery and retrofit
costs." (DX 13839, p. 2.)
CDC also found it difficult to establish a price for the
6600. In April 1964, CDC submitted 6600 proposals to the Bettis Atomic
Power Laboratory (BAPL) and the Knolls Atomic Power Laboratory (KAPL) ,
in competition with IBM, Burroughs, Philco and Sperry Rand proposals.*
(JX 10, U 12; Norris, Tr. 5620; DX 4960, p. 5.) The bidding process
^ i!
1 was highly competitive. Initially, BAPL and KAPL selected CDC and
IT!!
; Burroughs resDectively . Later, however, both BAPL and KAPL changed
IS
; their selections to IBM. (JX 10, fl 12.)
IS
;[ Six months later, CDC was told by the Government that it was
20
22
22
24
25 '
! * IBM proposed 360/90 's, with the interim installation of Model 70s
] until the 90s were ready for delivery. (JX 10, if 12.)
••{
.{
I -673-
L
Z
A-
4- ft
7
a
Iff
n
12
13
14-
interested in reopening the BAPL and KAPL negotiations if CDC was
prepared "to sharpen [its] pencils". (Norris, Tr. 5620.) According to
CDC Chairman Norris, BAPL and KAPL then misled CDC "in a deliberate
manner" as to the terms of the I3M offering, telling CDC "that IBM had
L I offered a computer at four times the power of the 6600 at a lower
Z\ price", as well as misrepresenting the date at which IBM could deliver
3 j its equipment. (Morris, Tr. 5970-73; PX 367, p. 5.) CDC made an
4- 1 "unsolicited proposal" to BAPL and KAPL in late February 1965 "at a
5- |. price substantially lower than that previously proposed by CDC and
5; substantially lower than the price proposed by IBM."* (JX 10, K 12;
7j DX 324.) CDC "proposed a combination deal which would involve replac-
ing the 6600 within some period of time . . . with the computer that
would be much more powerful than the 6600, . . . the 6800, and at the
time the 6800 was delivered, that we [CDC] would take back in trade
the 6600." (Norris, Tr. 5621.)**
CDC, "unfortunately" according to Norris, ultimately won the
BAPL and KAPL contracts. (Tr. 5963, 5976; JX 10, If 12.) Moreover, it
was unable to meet the delivery dates and, as a result, was required to
* CDC had earlier reduced the price on the 6600 because of "substan-
tial reductions in prices of component parts (transistors, diodes,
ijetc.) which . . .'occurred in [1963 and 1964]". (DX 13838, p. 4;
Schmidt, Tr. 27416.)
3
9-
10
11
IZ
13
14-
15
16
17 !
13 •.
ii ** One other aspect of CDC's pricing policies is worthy of note. Botifi
.q 'I Norris and Lacey testified that, as a general rule, CDC set the prices
it for its computer systems five to ten percent below IBM's prices.
2_ II (Norris, Tr. 5653; Lacey, Tr. 6567-70.) According to Lacey, this was
ji "[b]ecause our [CDC's] experience tells us that if we on a grand aver-
-- ;| age basis go significantly higher than that, that our opportunity for
^ j business rapidly diminishes". (Tr. 6573.) CDC considers the "prices
I of all manufacturers but principally the prices of IBM and, secondarily,
22 ij of other manufacturers". (Lacey, Tr. 6569.) Norris and Lacey admit-
-3 ij ted, however, that it is extremely difficult to compare accurately the
^ 1 performance of the system of one manufacturer as against that of another
; | manufacturer. (Norris, Tr. 6038-40; Lacey, Tr. 6800-01.) Norris
24 ij testified that computer companies compete on the basis of a variety of
!{ factors other than price; reliability is a factor, for example.
25 ii (Norris, Tr. 6040-41.)
; i
'!
II -674-
i
z\
i
31
i
?!
7 j
3
a
10
U
12
pay substantial penalties which further reduced its effective price.
The final settlement was "substantially disadvantageous to Control
Data". (Norris, Tr» 5976-78.) For a time, difficulties with the 6600
\ adversely affected CDG. According to Norris:
"We were losing money as a company in 1966/1967 primarily
because of problems with the 6600 computer. Frankly, there
was a great deal of conflict in top management in 1966 over
whether we should press forward or retrench — closing down
data centers was high on the list of retrenchment possi-
bilities. The decision was made to press on, however
there were some deserters in top management as a result —
they were afraid that the ship was sinking." (DX 284, p. 6;
see also Norris, Tr. 5678-79.)
Mr. Norris also testified that CDC "had to rush into the
6600" because it had been "literally clobbered by IBM competition" to
CDC's earlier 1604 computer system, and that with the 6600, CDC again
faced the "enormous impact of competition from IBM". (Tr. 5625.)*
f Ultimately, however, the 6600 — and the 6000 Series in general — proved
to be "particularly" successful for CDC. (See Norris, Tr. 5849-51.)**
In 1969, for example, CDC successfully bid for an Air Force procurement
ij to replace Univac- and IBM equipment with 12 or 13 6000 Series machines,.
17 j
i * Such militant language by persons speaking for firms in a competi-
IS ;j tive environment is not uncommon. For example, CDC's chief development
:j engineer for the 6000 Series, Vice President Seymour Cray, at CDC's
2Q j June 1963 corporate planning meeting, urged that CDC announce the 6600
J and a successor in order to "slug" IBM because, in his opinion, IBM had
21 ■! "made a mistake by putting all [its] eggs in an integrated circuit
!| basket". (DX 13526, Forrest, pp. 748-750.)
22 1
j ** CDC received more than $286 million 'in revenue and more than $185
23 j million in gross profits from the 6600 computer systems during the
[period 1964-1972. CDC's gross profits on the 6600 exceeded its gross
2*. 'j profit objective which was set to yield a reasonable rate of return on
j investment and a reasonable net profit at the bottom line. (DX 1185,
?s i pp. 3-5.) We are aware that DX 1185 is an offer of proof and not
1 evidence; however, we rely on this offer of proof because it is
;l consistent with the other evidence about the success of the 6600
:j systems and the growth of CDC.
I -675-
Li including several 6600s. The contract was for systems "to handle the
Z '■ entire inventory scheduling acquisition of spare parts for the Air
j jj Force Logistics Committee" and its aggregate value to CDC was approxi-
4. if mately $40 million. (Schmidt, Tr. 27469-76.)
CDC finally manufactured 94 6600/6700* computers (as compared
to some 17 360/90s manufactured by IBM, including four for use within
IBM) and a total of 215 CDC 6000 Series computers. (JX 10, 1f1f 35, 36.)
In the late 1960s, as a successor to the 6600 and a replace*-
ment for the never-delivered 6800, CDC developed the 7600 computer,
which it officially announced in December 1968 and first delivered the
following month — more than 21 months after the first committed delivery
date for a 6800 and seven months later than the delivery date called
for in the first contract using the machine designation "7600".
(Norris, Tr. 5628; JX 10, 1f 34; PX 355, p. 39.) Norris characterized
the 7600 as "several times more powerful than the 6600 and it addresses
the same market". (Tr. 5628.) CDC installed its first two 7600 compu-
ter systems during 1969 (DX 13843, p. 4) and in that same year, CDC
Vice President J. W. Lacey, speaking to a CDC graduate orientation
class, described CDC's success as follows:
3:
a
7 ;
a
9
a
x
,z
3
L*
^ ,
17 ;;
IS •
19 ii
;| " [W]e have a world-wide leading position in large computers
2C i today. That position is widely recognized. Since 1964, with
j the delivery of the first 6600 Computer, followed recently by
21 j the 7600 Computer, Control Data has dominated this market,
j Second, there is a rapidly increasing trend towards very
22 ;| large computers used in data processing networks in which
j many users share the enormous power of machines like the
22 ; 6600, and away from medium sized and small sized stand-
alone computers. ..." (DX 43 8, p. 7.)
24. }
i
25 |
! * The 6700, announced in May 1969, was "basically two 6600's".
j (PX 355, p. 37; Norris, Tr. 5626.)
j
; -676-
i
•i
•i
i
t
L
2
Z
c
7
a
10
n
iz
13.
14-
Looking to the future of the company, Lacey said " [w] e believe that
our position today and the direction we are giving our business puts us
in an outstanding posture to share in the explosive future growth of
our industry 1 ** ( Id. r p.. 13; see also Lacey, Tr. 6676-77.)
(ii) The 3000 Series > The 6000 Series was not the only
product line developed by CDC in the 1960s. CDC also significantly
expanded its 3000 Series. In September 1963/ CDC announced its 3200
computer; in January 1964, it announced the 3400. (PX 355, p. 34;
Norris, Tr. 5627.) Norris testified that the 3200 competed with "IBM.,
Univac, Burroughs, NCR to an extent, and possibly SDS". (Tr. 5627.)
1965 saw the continued expansion of CDC's 3000 Series.
The 3300 was announced in November 1965 and delivered in that same
month. (PX 355, p. 35.) Norris described the 3300 as CDC's "entry
into timesharing. And, again, I think it had some added features for
business data processing. And it was a considerably lower-priced
machine than, say, the 6600. It was what you term then a medium-
l[ size computer." (Tr. 5627-28.) According to Norris, it competed with.
| "IBM, Univac, Burroughs, SDS and NCR." ( Id. )
; The 3500 was announced in November 1965, although it was not
19 J
:| delivered until 1969. (PX 355, pp. 3 5-3 6.) Norris testified that it
2C !|
; was "essentially the same computer" as the 3300 except for the use of
21 i| .
:• integrated circuits and "somewhat larger memory options". (Tr. 5628.)
22 1
In 1967, CDC announced its 3150 computer, the smallest of the
22 i
i 3000 Series, stating that it "provides a complete business and scienti-
i
2- i
';} fic information handling capability with a minimum of hardware and
-677-
software. The. 3150 provides maximum throughput at low initial cost to
the user and the capability for him to expand upward as his information
handling needs grow.." (DX 13340, p. 8.)
The 3000 Series was a substantial success for CDC, without,
in large part, the start-up problems that beset the 6000 Series.
Indeed, as early as 1966, CDC was able to describe its 3200 — which was
introduced little more than two years earlier — as "highly successful" ►
(DX 13839, p. 5.) And in 1968 CDC reported to its stockholders that
"orders for our 3000 product line continue to increase-both in the numbejr
of systems ordered and in average dollar value." (DX 13842, p. 2.)
The 3000 Series was successfully marketed for applications in manufac-
turing, general business data processing, education, medicine, data
services and the brokerage business. (DX 13843, p. 4.) Moreover, it
gave CDC a lower-priced alternative to the expensive 6000 Series
computers.
b. CDC's Expansion into Commercial Data Processing. As the
:j decade began, CDC perceived itself as offering large, "scientific"
r ;{
; | computers. Very quickly, however, CDC learned that the distinction
H
: between scientific and commercial data processing — if indeed there ever
■j was on— had blurred almost to the vanishing point, and by the end of
* '< \
| the 1960s, CDC estimated that fullv 40 percent of its business came t
N - !
• from "pure business data processing". (Schmidt, Tr. 27476-78.) |
♦ CDC's Chairman Norris described CDC in the early and mid- ;
" 1 i
! 1960s as "a supplier of large-scale digital computers to scientific and !
i engineering applications". (Tr. 5624.) Norris also testified that, at
-678-
L
2
Z
4*
5
a
7
3
S-
LO
II
12
13
IS;!
17 I
IS
the time of its announcement in 1962, CDC thought that the 6600 "would
be unique to a great extent . • . it being so much more powerful and
so well-suited to scientific work, it would just be outstanding in the
eyes of those laboratories that have these very large scientific pro-
blems". (Tr. 5617-18.-)
CDC Vice President, Gordon Brown, described the entire 6000
Series as announced as
"very definitely a scientific line of computers, and there-
fore, the analyses that we did showed that the strength of
the 6000 product line prevailed . . . over IBM and Uni vac in
most typical environments; and, on the other hand, proved to
be deficient when it was employed in an environment requiring
a lot of input/output of data, or commercial type requirement
[because] the architecture of the 6000 series was designed
with the scientific user in mind. It had a large, fast,
central processor with a number of auxiliary processors to
handle the input/output functions. And it had a large, very
fast disk storage capability associated with it." (Tr. 50996.)
By October 1965, however, the CDC Executive Council* had
recognized that there were no longer separate markets for scientific
and business data processing. (DX 276; see Norris , Tr. 6002-06; Tr.
6081-82.) Thus, between 1964 and 1968, according to Brown, "gradually
additional capabilities were added to the 6000 computer system, and
these included COBOL compilers of sort and merge packages and the
ability to handle permanent files as opposed to using the input/output
2C ;
•i
devices as auxiliary storage or temporary storage of data files." (Tr. I
TT .1 1
;j 53064-65.) }
22 j . |
Similarly, CDC Vice President Schmidt testified that although ;
22 ; i
24 \
j * Lacey described the CDC Executive Council as "responsible for
25 j advising our Chief Executive Officer concerning major business ques-
tions". (Tr. 6556.)
-679- i
i
t
" [i]n the early stages of the 6000 marketing effort, we aimed at primar-'
ily the scientific applications . . . that has changed." The change
"started with the coding of a COBOL compiler for the 6000 Series and
the 6600 specifically. " (Tr. 27457.) CDC described the introduction
; of a COBOL compiler for its 6000 Series in 1967 as "an important
achievement, for we are now able to provide our customers with the full
I power of our super computers to handle the broad scope of their data
processing problems." (DX 13841/ p. 4.) Then came the development of
application programs for the 6000 Series in COBOL and the sale of "some
limited number of [business] applications, usually in conjunction with
primary scientific applications." By 1968, CDC had sold 6000 Series
systems in Mexico "which were devoted primarily to business data pro-
cessing, using that COBOL compiler and the COBOL application programs."
And in other situations, customers with business applications as well
as scientific applications ordered a 6600 system to do both. (Schmidt,
Tr. 27457-58.)
The primary impetus for the broadened use of the 6000 Series -
came from customers who wished to have a single machine capable of
performing both commercial and scientific applications — one of the
primary reasons that led IBM to develop with System/360 the capability
l to do both applications equally well. (Norris, Tr. 5618; see JX 38,
1 I •
.1 pp. 27-29; see pp. 290-96 above.) As Norris testified:
2 ;'»
j "We found that there were large companies who, while
3 the majority of the work that they wished to do was of an
engineering and scientific nature, still they had a certain
\i, j amount of business data processing and that they preferred
i to have only one computer as opposed to having two computers ,
?c i one for scientific and the other for business.
I
.1
1 -680-
i
L
2
Z
4>j
5
&
7
a
si
ia
tl i
i
12 i
13
15
"So, we set about to broaden out the software which was
available with the 6600 so that we could meet the requirements
of those customers where the bulk of the work was still scien-
tific but still the 6600 would do the business data processing
well enough so that the customer only had to have the one
computer. " (Tr. 5618 . )
While CDC also introduced the 3000 initially as "basically
scientific", it realized from the outset that the 3000 Series "had a
little bit more versatility as a business data processing machine than
the 6600". (Norris, Tr* 5627.) Over time, CDC added hardware features
and software packages to enhance the 3000 Series desirability for
business applications. (Brown, Tr. 50990-91.)
Among the uses of the 3200 during the first years of that
system's life were: medical research and training and use in "flight-
testing ground stations" (DX 13838, p. 7); and use in combination with
a 3600 "to integrate the computing and business data processing" for
the 57 associate companies of Phillips, the Dutch manufacturer. ( Id. ,
p. 9.)
• a ;! In 1965, CDC announced the 3300 which, according to Norris,
17 |( "had some added features for business data processing." (Tr. 5627;
IS jj PX 355, p. 35.) In its 1967 Annual Report, CDC stated that "the
IS it variety of applications being handled by the 3300 include production
20 ■•' scheduling, labor analysis, data communication, inventory control,
.1
ZL | engineering computations, and general business data processing."
22 1 (DX 13340, p. 8.) And by 1966, according. to Brown, "the 3000 product
23 i line . . . was evolving to a . . . better balanced product line between
24 j both the scientific and the commercial users. The initial base of
.!
25 I customers had largely been scientific users, and many of them were
■I
;i -631-
1
2
Z
4- [GDC did not manufacture- its t own peripheral equipment. (Norris, Tr.
J
6
7
i
i
s;
!
?!
i
u j
12 j
13 I
starting to expand their applications for commercial usage." (Tr.
50-9-97/ see Tr... 50990.)
c. CDG's Expanding Peripheral Business. In its early years
5609r see PX 6066, p. 1.) In the 1963-1969 period, however, CDC began
to manufacture peripheral equipment not only for attachment to its own
processors but also as an OEM supplier for other EDP companies. In
addition, CDC laid the foundation for its later very successful entry
into the IBM plug compatible peripherals business.
By the early 1960s, CDC had recognized that the sale of
peripheral equipment was potentially a highly profitable opportunity
and therefore began to expand its peripheral offerings. For example,
in 1964, in its news release announcing the 6000 Series, Norris stated:
t4. | . " [Nlumerous peripheral devices . . . are under development
and will be announced over the next two years to complete
the implementation of products required for total manage-
ment information systems. These peripheral devices include:
15
16
17
IS
19
20
;l
-, :! "Optical character recognition readers.
"Line of visual displays.
"Disk files — not only low-cost units but very sophisticated,
high capacity, low access time, extremely high transfer rate,
mass memories.
it "Mass core memory.
"Remote terminals and processors for on-line man/machine
interaction.
22
23*
24
25
"Line printers, card ounches and readers."
(DX 319, p. 3.) j
In 1965, CDC acquired Data Display, Inc., a manufacturer of j
"electronic display peripheral equipment". (PX 355, p. 5; DX 296.) In j
I
I
-682- i
1965, CDC announced its 852 disk drive which was "in many ways like the
IBM 1311". It was marketed "in a very modest way" on an OEM basis to
GE and Honeywell in Europe (Brown, Tr. 51015-17) because "there was
very little market demand for this type of product at the time, and CDC
was just beginning to build and staff an effective market and service
organization." (Brown, Tr. 51015-51017.)
In 1966, CDC announced its 9433/34 disk drive — an IBM 2311-
type device, although it was not media compatible with the 2311 drive —
on an OEM basis, with first shipments occurring in 1967. CDC's princi-
pal OEM customers for the 9433/34 were Honeywell, GE and RCA, as well
as ICL in Great Britain and Siemens in Germany.* CDC eventually sold
some 16,000 94 33/34 drives in the late 1960s on an OEM basis, at prices
less than one-half that of the IBM 2311. (Brown, Tr. 51056-58.) The
development and marketing of such IBM-type devices, of course, fore-
shadowed CDC's later decision to produce IBM plug-compatible peri-
pherals. (See Brown, Tr. 51063-67.)
In the 1966-67 time frame CDC also began to market peripherals
I originally designed for its 6000 Series on an OEM basis, such as the
it
; 6638 disk file, which was sold to Honeywell, ICL and GE as the 94 90.**
22
23
24
25
"We view the computer market as a world market , and plan our
organization and operations to maximize our abilities to best
satisfy our customers in that market." (DX 14214, P- 5; emphasis
in original.)
** Less than 40 9490 disk files were shipped, however, according to
Brown, because there was not a large market at that time for that type
of fixed non-removable storage disk device and because not many OEM
customers had channels that could take the high data rate of the 9490.
(Tr. 51033-34.)
-683-
}
r !
(Brown, Tr. 51032-34.) In 1967, CDC introduced a new line printer, new
tape transports, a card read-punch, a magnetic drum storage unit and
several new versions of electronic display terminals. (DX 13840, p.
8.)
During 1968, CDC added to its peripherals product line "a 5
billion bit disk file, a 1200 line per minute printer, and a new gene—
r i| ration of tape transports"- (DX 13841, p. 8.) At the same time, it
informed its stockholders that " [independent suppliers and the in-
house developments of major computer manufacturers do and can be
expected to continue to intensify competition." ( Id . , p. 8.)
And in 1969, CDC introduced six more new peripheral products:
a disk storage unit, two printers, a card reader, a display terminal,
m -\ and a drum device. (DX 13843, p. 6.) Also, CDC announced an IBM 2314-
.type device in 1969 for use with CDC ' s 3000 Series, 6000 Series and
CYBER 70 product lines. (Brown, Tr. 51068-69.)
By the end of the 1960s, CDC "had made major investments in
5 ;!
;[ technology in most of the principal peripheral areas. This started
7i|
j with the development of subsystems for use in [CDC's] own computers and
■ carried through most of the Sixties . . . into the development of a
- j j
5 fairly large base of OEM business." (Brown, Tr. 51212; DX 4 38, p. 12.) j
j
j d. Data Centers. CDC also greatly expanded the data center i
2. 1 - !
\ (service bureau) portion of its business in the 1963-1969 period. In j
'2. :S i
~ 11964, for example, CDC had six data centers. (DX 13838, p. 8.) By j
"* !l969, it had more than 40 "throughout the world". (DX 13343, p. 6.) j
I The six data centers oaerated bv CDC as of 1964 used CDC 36 00
.( *■ ■* i
?e ■ !
~* I !
I j
:j -684- I
;■ i
"» i
i I
L
2
3
4
5
S
7
3
&
10
II
12
13
14
15
and 1604-A computer systems, forming a network — later known as
CYBERNET — "tied together by Bell System Data-phones" and "providing
complete data processing services to commercial and government users oi
a contract basis." (DX 284, p. 4; DX 13338, p. 8.) Typical of the
many applications processed at the centers were "Operations Research
applications", "traffic surveying and planning", "Hospital data
processing" and "school scheduling and grade reporting". (DX 13838, p,
8.)
By 1965, CDC had seven data centers and had begun its "net-
work development". (DX 284, p. 4.) However, "[e]xcept for brief
periods in the mid-6 0's, data centers in the aggregate operated at a
loss until 1972 because [CDC] kept pouring money into expansion".
( Id. ) * Norris, in a draft of a speech in 1973, cited this as an
example of CDC's "willingness to take risks". ( Id. )
In the fiscal year ending June 30, 1968, CDC acquired C-E-I-I
•g jia company which offered computer programming and other professional
data processing services, and Pacific Technical Analysts, Inc., claimed
to be "the largest and most capable programming and service center
'{company serving the Western Pacific area". (DX 13841, p. 2.) By the
j end of fiscal 1968, CDC was operating over 30 data centers worldwide
i
17
18
21 !
22
23
24
25
and offering "an extensive inventory" of application programs. (DX
13841, p. 7.)
In 196 9, CDC offered the following description of its
CYBERNET network of data centers:
* Accounting losses which result from expenditures made for the pur-
pose of achieving later returns, are, of course, not truly economic
losses but rather investments.
-685-
I i
i
z\
I
5-1
e
7
8
9-
L0
n
L2i
13
14
15
16
"Through the highly advanced CYBERNET service, customers
have convenient access to the cost /performance advantages
offered by both the CDC 3300 and 6600 computers without having
to make large capital outlays." (DX 13843, p. 6.)
With more than 13,000 miles of communications lines by the end of
fiscal 1969, CDC was offering its data services at more than 40 data
centers throughout the world, a more than 550 percent increase over its
1964 holdings. ( Id. )
e. CDC's Acquisitions (1963^1969) . The story of CDC's
expansion in the 1960s cannot be fully understood without considering
CDC's acquisitions during that period. Between 1963 and 1969, CDC
acquired some 43 companies, at a total cost of over $897,000,000.
(Norris, Tr. 5788-89; PX 355, pp. 3-9; DX 296.)
All of those companies — with the exception of Cedar Engineering,
Kerotest and Commercial Credit Corporation (which is discussed in some
detail below) — were supplying an EDP product or service at the time of
acquisition. (Norris, Tr. 5794-95.)
CDC's numerous acquisitions, most of which were paid for by
17;JCDC stock (Norris, Tr. 5789), enabled it to broaden its product and
»
18 ; service offerings quite rapidly without the substantial development
!
19 ; time that internal expansion would have required.* As Norris stated in
2G :;a draft of a 1973 speech:
n
T ii
22 jj * That is not to say that CDC did not expand internally as well,
I particularly through increased vertical integration. For example, CDC
23 ij decided in 1966 to, have its research division manufacture integrated
;j circuits for use in the prototype of the 3500 computer rather than buy
24. Ii circuits from Texas Instruments. (DX 432.) Also in 1966, CDC reduced
;• costs by bringing the manufacture of card module assemblies, memory
25 J cores, memory planes, memory stack assemblies and logic chassis assem-
blies in-house. (DX 13339, p. 9.)
'i -686-
"Our high P/E ratio stock, or Chinese money, as
we often termed it, was used to acquire companies with
complementary technology, products, services and markets.
In other words, we were not trying to broaden our base
as in a conglomerate, but rather to buy new computer
products and services and markets to spread development
costs- and gain economies of scale as rapidly as possible."
(DX 284, p. 7.)
Norris agreed that this was "an alternative to investing money in
research and development" and was "very successful for Control Data for
that purpose". (Tr. 5804-07.)
Speaking about the acquisitions in general, Norris testified:
" [W] e wanted to have our own peripheral equipment
to put on our computer systems so that we would have full
control over the cost and quality. We wanted to broaden
out our product line both with respect to hardware as
well as software.
"In some instances we bought data services businesses,
which gave us additional revenue and profit. And we were
able to take those services in turn and have them sold by
a larger marketing organization." (Tr. 6092-93.)
Thus, in the fiscal year ending June 30, 1964, "Control Data
made significant additions to its technical capabilities and product
lines, and broadened its market areas" by way of a number of acquisi-
tions. It acquired companies with capabilities in the areas of:
digital computers for use in power, chemical, petroleum and oil indus-
is i
:[ tries; card punch and reader systems and other peripheral devices;
2G !
! optical character recognition equipment; data collection systems,- data
21 : i
:! processing services; printers; and analog to digital conversion equip-
22. ">
■j ment. (DX 13838/ p. 5.)
22 1
j In the fiscal year ended June 30, 1965, CDC acquired
24, :\
j companies with capabilities in electronic display devices and pro-
2 C i
j gramming consulting services, as well as a business data processing
i
i
1
I -687-
i
2 ;
i
i
e
7
a
9-
12.
12
14-
15
16 i|
17 ;!
is
1
Z
3
4-
5
t\
sj
9 I
ia j
i
i
11 :
13
IS
center and two companies whose products, involving radar, for example,
incorporated the use of digital computers. (PX 355, pp. 5-6.)
In the year ended June 30, 1966, CDC acquired the commercial
computer operation of General Precision Inc.'s Librascope Group.
"Included in the purchase were General Precision's commercial computer
rental and service contracts, and inventory of commercial computers" as
well as its "highly experienced commercial computer sales and service
organization. " Other acquisitions were of an electronic systems
engineering company, a Hong Kong firm doing assembly of electronic
components, particularly ferrite cores, and an Italian firm operating
data centers in Italy. (DX 13839* p. 3.)
In 1967, CDC acquired, among others, C-E-I-R, (DX 296, p. 2)
because, as Norris testified:
" [C-E-I-R] had the American Research Bureau, which
was using a computer in surveying the listener response
in the television and radio industries; there was Automation
Institute — these are schools to teach computer programming
and computer operation; there was a data services business
IS ;j .... And it was primarily those three areas that were
particularly interesting to Control Data." (Tr. 5796-97.)
17
IS
19 j
CDC Vice President Lacey testified that the acquisition of C-E-I-R "was
an additional entry for Control Data into the data services and consult-
i
I
ing services business, beginning steps of our broadening or our business!
J line". (Tr. 6632-33.) I
j CDC's single most important acquisition — Commercial Credit —
j occurred in 1968. (PX 355, p. 3.) In August 1968 CDC acquired
43 i
! Commercial Credit, "a diversified financial institution . . .
2*1
| with nationwide and Canadian operations in financing, lending, leasing,
25 !
'.factoring, and insuring" (DX 13342 , p. 16) for 4,825,720 shares of CDC
-688-
stock, with a total market value of $745,573,740 (PX 355, p. 8) — by far
the most expensive acquisition made in the 1960s by Control Data.
( Id. t PP» 3-9.) The principal reason CDC acquired Commercial Credit
■■ was to gain a financial services subsidiary in order to enable CDC
better "to finance computer leasing." (Norris, Tr. 5643; see Lacey,
Tr. 6586-88.)
Initially, CDC marketed its computer systems on a purchase-
only basis. However, by 1961 or 1962, CDC had realized that many EDP
customers demanded leases and, accordingly, it began to offer its
system for lease as well as purchase. (Norris, Tr. 5641-42.) Hence,
CDC over time has offered one-year, three-year, five-year and longer
leases. (Norris, Tr. 5644.) CDC first offered three-year leases in
1966 and non-cancellable five-year leases in 1967, both at a discount
from its short term lease price. (DX 295.)
CDC's changeover to leasing as well as purchase required that
|j CDC "raise additional sources of capital . . . [ijn order to finance
the leases [because] when you lease a computer you get paid on a
monthly basis , but you have to incur the total cost of the computer at
the time it is delivered." (Norris, Tr. 5642-43.) Thus, in 1966, CDC
L
Z
z,
4>
a
6
7
3
ia
u
12
13
14-
15
15
17
IS
19 ;i
it entered into an arrangement with Leasco whereby Leasco would purchase
20 :
| CDC systems and then lease them to customers on a long-term basis. (DX
21 "\ •
j 13839, p. 2.) However, CDC cancelled that agreement the following year
22 "»
j "in light of current and prospective financing plans of the Company".
23 j
(DX 133*0/ p. 12.) According to Norris, it was not until CDC acquired
2*1
:j Commercial Credit Corporation in 196 8 that CDC ultimately "solved the
25 '■!
[problem of financing leases". (Tr. 5643-44.)
-689-
f.. Conclusion. When the 1960s began, CDC was a virtual new-
entrant in the EDP industry, having been incorporated only three years
earlier* (DX 271,, p. 7.) It perceived, itself as being principally, if
: not solely, a supplier of large scientific computers. (Norris , Tr.
5624,) By the end of the 1960s, CDC was firmly established as a major,
diversified competitor in the EDP marketplace. It had achieved great
success with its 6000 Series computers. It had added considerably to
the business data processing capabilities of its computer offerings , to
the point that by 1969 approximately 40 percent of its business came
from "pure business data processing". (Schmidt, Tr. 27477-78.) It
had greatly expanded its peripheral equipment offerings and begun a
successful business as an OEM supplier of peripherals. And it had
increased the number of its Data Centers from six to 40. (DX 13838, p.
8; DX 13843, p. 6..) The 1960s were indisputably a period of great
success for Control Data Corporation.
»!
■\
i
\
":i
- '!
-690-
z
3.
5"
5
7
3
9
ia
12
13
4-7. SDS .. Scientific Data Systems ("SDS") was formed in
1961 with an initial capitalization of approximately $1 million,
raised from a San Francisco venture capital company and the firm's
original founders. (Palevsky, Tr. 3128, 3193*)* SDS was the idea of
Max Palevsky who furnished approximately $60,000 to $80,000 of its
initial capitalization: "I put up half the money in cash and half as
a note." (Palevsky, Tr. 3127, 3193.) For his investment, Palevsky
received "something in excess of 15%" of SDS * s equity. (Palevsky,
Tr. 3193-94.)
Palevsky had begun his business career as a research analyst
for the Bendix Corporation in 1952, in "the division of Bendix that
was starting to explore computers". His responsibility was
"[plrimarily logic design, computer design .... It was a time when
T £ :■ everybody did everything". Upon leaving Bendix in 1956, Palevsky
.- : organized the Packard Bell Computer Corporation, a subsidiary of
. £ ! Packard Bell, which "buil[t] specialized digital computers, special
purpose digital computers, and eventually a small general purpose
u
iS ;i
} * The witnesses testifying about SDS/XDS (SDS was called XDS —
i - ;[ Xerox Data Systems — when it became a division of Xerox) were Max
| Palevsky (described above) , Harvey Cohen and F. R. Currie. Only
23 j Palevsky was with SDS in the early years. Cohen arrived about 1964
i and held a number of positions, including in 1967-68 the Director
21 | of Marketing Operations. (Cohen, Tr. 14427-28.) Currie also came
-j in 1964 and held various marketing positions, becoming Vice-President
22- : ] of Sales in 1968-69. (Currie, Tr. 14909-13.) At Xerox Data Systems,
•j Cohen became Vice President of Advanced Systems, Business Development
23. j Group (Cohen, Tr. 14427-28, 14521) and Currie became Vice President
j of the Data Processing Division reporting to Cohen. (Currie, Tr.
2- 'I 14917, 14922-23.) Currie later moved to the Corporate Marketing
! staff. (Currie, Tr. 14923-24.)
i -691-
•t
1 !, computer and digital systems".* (Palevsky, Tr. 3121.) In 1961,
2 : Palevsky left Packard Bell because
3 ;| "that company had come on hard times. The ideas I had about how
J to proceed in the computer industry required much stronger
4 j| backing from the parent company which they could not provide.
I also felt that the computer industry is a very unique
51; kind of industry, and it was very difficult working under a
|| management that really knew nothing about the industry itself, so
5 jj that it made sense to be independent, and, of course, there wera-
jj opportunities to make a great deal of money". (Palevsky, Tr.
7 jj 3127-28.)
•i
8
9
10
11
SDS initially conducted all its activities in a 5,000
square foot facility with approximately 17 people of whom 12 were
professionals. (Palevsky, Tr. 3196, 3198; DX 45, p. 4.) Its first
product was the SDS 910 computer system, delivered in mid-1962, less
.- jj than a year after its organization. (PX 5774, p. 13.) That first
j2 jj product was designed to take advantage of an opportunity perceived by
, . ;j SDS for high performance hardware offered with little support to
i g !i sophisticated customers for use in real time applications . The
.i
.,. \ market opportunity in fact existed and the 910 and subsequent
lo
i! products were very successful. SDS built on the success of its
; ; initial specialization. Throughout the 1960s, it successfully
;i expanded its product line both by offering its computers to a wider
: set of customers for a wider range of applications and by producing
: more and more of its own peripherals and software, which it had
;; previously acquired from other vendors.
22 "
: * Palevsky defined a general purpose digital computer as "an
*** ; electronic device with a stored program, which may be changed, and
; depending upon the program, can operate a large variety of tasks".
d5 , (Tr. 3132-33.)
: ' -692-
L
2
3
4-
f
e
7
3
10
IX
12
12
14*
15
10
17
IS
19
SDS grew at an extraordinary rate while also achieving
substantial profitability — in fact, SDS "produced continually increas-
ing profits virtually from inception". (PX 5774, p.. 6.) Its average
annual compound growth rate from 1962 to 1968 was 115%. Even after
the first two years, it continued to grow at a rate of approximately
50% per year. (DX 46, summarizing data contained in DX 44 and DX
45.) Its revenues, which by 1964 had reached $20.5 million, rose to
$100.7 million by 1968, the last full year before it merged with
Xerox. SDS was merged with Xerox in 1969, in exchange for Xerox
stock valued at approximately $980 million. Of this amount, Palevsky
received approximately $100 million worth of stock. He had also
received several million dollars from previous sales of SDS stock.
(Palevsky, Tr. 3195-96.)
a. The SDS Entry Strategy. SDS implemented a consciously
determined strategy to capitalize on what it saw as a market oppor-
tunity. Palevsky testified that at its formation in 1961, SDS had
"two markets" in mind for its products, "one market being what I
would characterize as the real time computer market, and the other,
the small to medium scientific computer market". (Palevsky, Tr.
^ J 3133.)* SDS began to market computers of high performance hardware
offered for real time applications to customers that did not need a
21
22
* Palevsky described "scientific data processing" as processing
where "a relatively small amount of data is entered into a computer,
a large number of arithmetic operations are performed on data and a
relatively small amount of data is produced in some printed form".
"Business data processing", he said, "has the opposite meaning".
(Palevsky, Tr. 3136.)
-693-
21'
3
A
5
>5
.7
lot of software and support services from the manufacturer.*
(Palevsky, Tr. 3137; PX 2103 (Tr. 23290).)
As; Palevsky testified:
"It was; p»rt of, the- market, that essentially no one
had attended to . At thdt time the other companies were
really concentrating primarily on computers as devices
into which one fed documents that contained data, cards,
tapes, etc., and out of which one got printed answers.
"Our computers were, intended for a market which fed
real time data, that is, data that came from centers in a
steam generating plant or a missile launching site or some
astronomical instrument and produced signals that, say,
worked the valves on a steam generating plant or indicated
to other pieces of equipment within the launch site the
status of various functions within a space vehicle so that
it didn.' t work as a computer works in an air-conditioned
computing center, but rather as part of the whole complex
of operational equipment". (Tr. 3135.)**
This strategy was highly successful. Palevsky testified
that SDS , "at the beginning", was able to sell its products with "a
very large gross profit" :
"We were able to do that at the beginning because we provided
hardware, that range of hardware and other services that was
SDS did not initially attempt to market its product to "business
.8 ■] data processing customers" because SDS "didn't have the kind of
^| people who understood the business market and the need of the
L9 i business market and we had not developed the software, the appli-
1 cations engineering, the general support that the customer needed".
iO !j (Palevsky, Tr. 3137.)
i
21 :| . ** In its 1966 Annual Report, SDS described its formation as follows
;i
22:1 "In 1961, when SDS was founded, highly experienced technical
j personnel skilled in the design, production and marketing of
23 [ small scientific and systems computers were uniquely available
in southern California. During that same period, the scientific
and engineering segment of the computer market required small,
I real-time computers which could monitor and control experiments
25 | or testing programs and rapidly process the results. Recog-
1 nizing the requirements of this market, the initial objectives
•j of SDS management were to attract competent technical personnel
\ and effectively apply their experience to meet this demand."
; ; (DX 982, p. 4.)
I -694-
L
Z
3
4.
ff
7
3
?
IQ
LI
12
13
1*
Iff
relatively unique and consequently the customer was willing to
pay a relatively large sum for it." (Palevsky, Tr. 3155.)
DEC was the only other firm Palevsky remembered producing products
similar to those of SDS in the early 1960s. (Palevsky, Tr.. 3136.)
The recognition by users of the potential benefits of the
early SDS products — obvious from SDS ' s growing revenues — was expressed
by NASA's Ames Research Center, which described its procurement of an
SDS 920 computer system in 1963 as follows:
"The integration of digital computers into physical systems
dedicated to specific areas of research has only recently become
economical and feasible through the reduction of equipment costs
and component size. One of the first such systems in use at
Ames was installed in 1963. This computer (SDS-920) was pur-
chased by the Guidance and Control Systems Branch and applied to
research on on-board computer and display requirements for
spacecraft and aircraft*" (DX 5316, p. 9.)
Subsequently, and rapidly, SDS expanded its product line and
its marketing approach. SDS attributed its "early and sustained
profitability" to its ability to meet the needs of its users:
t ^ ., "Because of the rapid growth of the computer industry, the
ic |j age of a company has not been a principal factor in its profes-
:| sional or financial maturity. Far more critical in a company's
-' »( potential is its ability to understand and act upon the changing
' requirements of the marketplace. It is to this posture that SDS
- s ;| conforms." (DX 44, p. 5.)
i3 ;j b. The SDS 910. SDS ' s first computer was the SDS 910,
2C ,} which Palevsky described at trial as a "special purpose general
.1
21 :l purpose computer" — by which he meant "a computer that had all the
i
22. :! characteristics of what was generally known as a general purpose
23 ;i computer, with the added capability of operating . . . within a
24. S systems environment, that is, it was a computer that was easy to
i
25 i integrate with diverse types of special purpose equipment". (Palevsky,
i
1 -695-
Tr. 3132, 3134.) The "main frame" sold for $80,000 to $90,000?
"[t]hen, depending on the peripherals, it got more expensive"-.
Palevsky described the peripherals as being, "at the beginning,
rather primitive equipment": paper tape punches, paper tape readers
and card equipment. Also, at the beginning, the SDS 910 was marketed
with "very primitive software, really just an operating system".
(Palevsky, Tr. 3134,)
c. The Expansion of the SDS 900 Series. In 1963, SDS had
expanded its line by announcing the SDS 920, 930 and 9300. (DX 44,
p. 7.) Those systems were compatible with the 910 and were designed
with a "building block" design philosophy. ( Id. ) The 92 and 925 were
introduced in 1964.* ( Id. , p. 3.) In its 1964 Annual Report, SDS
told its stockholders that with the introduction of the "small, high-
speed SDS 92 and the medium scale SDS 925, the company now offers a
family of six compatible, general purpose computers — the SDS 92, 910,
920, 925, 930 and 9300 — providing the flexibility required for both
i| industrial and scientific systems". ( Id. , p. 7.)
* SDS did not actually "manufacture" its 900 series com-
ii
li puters; rather it assembled them. That is, SDS purchased the various
25
* "The 925 was a modification of the 930 to . . . provide a faster
lower-priced machine. It was software compatible, and was really
just a modification of another product." (Palevsky, Tr. 3214.)
-696-
L I — Certain basic components for its central processing
i
Z '}■ units and memory (i.e., transistors, resistors and capacitors)
3 I from the "fs]tandard avenues of supply — [f]rom the manufacturers
4*1* of those components". (Palevsky, Tr. 3198-200.)
3 \\ — Gore memories "at the beginning from Fabri-Tek" , which
5 It was "one of a number of companies that supplied core memories";
7 S{ SDS acquired the memories in the form of core stacks, then
1
3 !| assembled them in boxes. (Palevsky r Tr. 3199.) SDS subsequently
acquired core memories from Ampex, Magnetic Memories "and
probably one or two others"; when "we got to a certain size we
generally had three sources of supply so that we were always
assured that one of them would be there". (Palevsky, Tr. 3200-
01.)
— Tape drives and tape control units from Ampex, Computer
Froducts, and Potter Instruments; eventually, SDS made its own
tape drives and controllers. (Palevsky, Tr. 3201-02.)
— Printers and a few printer control units from NCR and
Data Products and, in the case of "some specialized ones", from
"small companies"; at the time of its acquisition by Xerox in
1969, SDS was buying printer mechanisms from NCR. (Palevsky,
I Tr. 3202; see Plaintiff's Admissions, Set I, 9 KH 191. 0-. 2.)
i
1 •
J — Disk drives and disk drive controllers from Control Data
(and "perhaps some of them from California Computer Products") .
; (Palevsky, Tr. 320 3.)
2± 1
i
"* i
i -697-
9
ia
LI
12
12
14 ;
15
16 t
is!
»
21 ;t
22
z
1
,4
5
J J
L9
— Card punches from Univac; card readers were initially
acquired from a third party, but SDS subsequently built them
itself. (Palevsky, Tr. 3203.)
— k, few cathode ray tube terminals from Control Data;
however , SDS built most of these itself. (Palevsky, Tr. 3203.)
— " [Sltan&ard [Teletype] keyboard devices from Western
Electric " . ( Palevsky , Tr * 3204.)
SDS both wrote software for its computer systems and used
outside software services. Software services were provided by
Programmatic s and another firm, and "a number of smaller firms for
very specialized things". SDS also had "a number of users groups and
a number of our users ' programs became standard programs that were
then widely distributed" by SDS. Additional software was obtained
from a European company (a predecessor of CII) , which was licensed by
SDS "to build our computers in France". (Palevsky, Tr. 3205-06.)
SDS itself furnished maintenance service. (Palevsky, Tr. 3134-35.)
The SDS 920 had certain instructions that were not included
in the earlier 910 and "a slightly more sophisticated input-output
system". It was marketed to essentially the same customers as the
910. (Palevsky, Tr. 3162.)
;| The SDS 930 was "larger and faster and, again, somewhat
11 \
\ more complex structurally". It was partially marketed to the same
21 ■:;
j group of customers as the 910 and the 920, but was also marketed "to
23 I
S a greater extent to the general scientific community". (Palevsky,
U I
1 Tr. 3162.) For example, an SDS 930 was used for data reduction and
mm
i
} -698-
L
2
3
4-
3
7
a
71
12
13
14
15 ;
IS
analysis at the Mississippi Test operations center associated with
the NASA Slidell Computer Complex (DX 5836, Reeves, pp. 55-56) and,
at the Kennedy Space Center, an SDS 930 performed off-line simulation
of launch vehicle events for training, supplied input data to Mission
Control in Houston and handled a "fuel loading" system. (DX 5652,
pp. 116, 123, 164.) Palevsky testified that when it was first
introduced, the SDS 930 competed with the IBM 1620 "[a]nd then when
the 360 was introduced, the 360/30s, 40s and 44s". (Palevsky, Tr.
3185.) Other competitors with the 930 were Computer Control Company
(later bought by Honeywell) and DEC. (Palevsky, Tr. 3165.)
Palevsky described the SDS 9300 as "conceived much more as
a data processing system, as a computer that would sit in a central
computing facility and essentially provide printed answers, as
opposed to being interconnected on a real time basis with other
sources of data". It was marketed "to the scientific community", but
performed a still broader mix of applications. One customer was
,_ ; DuPont, which had' previously integrated SDS computers "into systems
:■[ for controlling chemical processes". It acquired this new computer
I not only for a "specific process they wanted to control but rather
19 i
! for a general computing purpose, so that the customer may have been
20 i
J the same, but the part of the company would be different". (Palevsky,
f i
i
! Tr. 3163.) Similarly, Digicon, Inc., used a 9300 to process seismic
22. '*>
t data collected from oil fields and also to process its accounting
I records. (DX 4085, Poe, pp. 18-19, 21.)
2± : )
! Palevsky testified that with the 9300, SDS "had now entered
25 S
. -699-
i
r
3 !
.if.
7
3
9
a
3
L5
La
17 !
the more traditional and more highly developed market for computers
and no longer had the edge of the innovations" it had made in "real
time computers" on which to rely.* This engendered "[a] 11 the
difficulties that go with, a highly competitive sales situation —
marketing situation" \ (Palevsky, Tr.. 3164.) Palevsky testified that
SDS's "main competitors' 1 * in marketing the 9300 were IBM and Control
Data (and later, Digital Equipment). (Palevsky, Tr. 3165.)
SDS had supported a "growing program of research and
development" and committed "substantial capital to advanced product
planning". One of the results was its announcement in 1964 of what
it claimed was "the first computer to use monolithic integrated
circuits, the SDS 92". As a result of the use of integrated circuits,
SDS's manufacturing costs were "decreased while the reliability of
SDS computers is improved at least three times over present models".
(DX 44, p. 5.) Withington, writing in 1964, concluded that:
"The most significant development in components has been the
approximately 50% reduction in the manufacturing cost of high-
speed circuits over the past three years . This quite rapid
development has enabled new small companies (e.g., Scientific
Data Systems, Digital Equipment Corporation) to enter the com-
-3 ; { puter market with low-priced computers of high performance. . . .
|j This reduction in manufacturing cost has been at least partly
IS \ responsible for the recent price reductions on older computers
! and the lower prices of new ones. The user has benefited, and
2G j the market has been enhanced." (PX 4829, p. 31.)
'I
21 j By the end of 1964, SDS told its stockholders that its com-
i
22 ; puter systems were "presently being used by industrial, scientific and
£-3 '
24. I * "As the technology in the computer industry evolved, there were
i no longer those pockets , there were no longer those market areas that
25 j had relatively little competition." (Palevsky, Tr. 3155-56.)
j
"] -700-
L
2
government organizations in many diverse applications ranging from
space exploration to construction, medical research to food process-
ing". (DX 44, p. 3.) The number of SDS employees had increased from
438 in 1963 to 1,357 at the end of 1964. ( Id. ) The year 1964 was
also a year of expansion abroad. According to SDS's Annual Report for
that year, " [f]rom nuclear experimentation in Geneva to automotive
manufacturing in Tokyo, SDS computer systems are finding an accelerat-
ing and receptive market throughout the world. . . . In the first
significant year of SDS activity abroad, computers were ordered or
installed in more than 15 countries." ( Id. , p. 17.)
Expansion, plus SDS's program of research and development,
required capital. Requirements for capital also increased because,
according to Palevsky, SDS was leasing more of the 9300s than it had
prior computers. (Palevsky, Tr. 3164.) In 1964, "due to [its] rapid
growth", SDS made its first public offering of common stock, offering
Tg \ 382,375 shares and raising almost $5,000,000. (DX 44, pp. 3, 21.)
SDS engaged in a continued pursuit of growth and expansion
through continued product improvements. In SDS's 1965 Annual Report,
1Q :! following the announcement of IBM's System/360, SDS stated:
•t
;| "The character of the computer market changed substantially
^ j last year as the result of advances in both the understanding of
the technology and in the manner in which computers should be
employed. . . . [T]hese changes point to the increasing use of
4
4-1
5
7
3
9
10
n
12
i
14- 1
15 I
17
■\
18
7f
22 *
I
23 1
| total management information systems for business, scientific,
aerospace and industrial control applications.
"As is always the case in the computer field, the new
market demands increased performance economically, in terms
2- ; of more computations per dollar, and an expanded array of
supporting services such as programming, field services and
/*s !
^ \ training." (DX 981, p. 4.)
-701-
L
2
3
4-
3
In 1965, SDS brought out the 940, designed for simultaneous
access by multiple users at remote locations. SDS took the 930,
increased its memory capacity, and integrated rapid access data
storage units and communications equipment. A 9 30 system costing
$250,000 was thereby transformed into a 940 system costing $1 million.
(DX 981, p. 3; DX 982, p. 12.) SDS called the 940 a "timesharing
7 |l computer". It was used, among others, by several commercial time-
sharing service bureaus (DX 45, p. 7) as well as by a data center
established by SDS itself to sell time to remote users in the Los
Angeles area. (DX 983, p. 3.) Similarly, White, Weld & Co. used its
940 to implement a financial information system that permitted its
individual subscribers to request portfolio information on a variety
of companies. (DX 982, p. 12.)
By the time of the introduction of the 940, SDS had announced
the development of its own magnetic tape units and rapid access disk
.£ ; i files as well as a line of digital logic modules. (DX 981, p. 3.)
,„ '} Also, in mid-1965, SDS "announced a new business programming*
it package for all its computers to supplement the extensive library of
- s ;l
.- !{ programs presently available to scientific users". The package, known
i
;t as MANAGE, was "expressly designed to facilitate corporate decision
20 j
l making bv management personnel outside of the data processing depart -
21 ■! •
|i ment" . (DX 981, p. 4.) Similarly, SDS adopted some of the marketing
22. 'I
\ practices of others in the industry. In 1965-66, it offered the
23 i
1 Federal government a 14% discount towards equipment purchase " [f]or
2* 1
! qualified Government schools and training institutions when primary
25 :j
i
-702-
3
10
II
12.
12
14-
15
I
2
3
4-
5
a
7 ;
a
9 ;
ia
II
application of the data processing system is for educational and
training purposes". (DX 47, p. 16.) It also offered the government
the provision of "programming aids , including programs [ , 3 routines ,
sub-routines, translation compilers and related items without extra
charge". ( Id. / p. 3.) By 1966 SDS was marketing its computers on a
variety of lease terms as well as selling them. (Palevsky, Tr. 3207.)
SDS continued to grow dramatically during 1965. It doubled
its number of installations in one year. New business received during
the fourth quarter of 1965 was greater than any previous quarter in
the company's history. To finance this continued expansion, SDS sold
47,500 shares of $100 par cumulative preferred stock to a group of
y* j insurance companies, raising $4,750,000 in the second quarter of 1965.
22 | And in February 1966, it sold $10 million in convertible subordinated
• . j debentures to a group of institutional investors. (DX 981, p. 3.)
d. The Sigma Series. By 1965 SDS had begun the develop-
ment of its Sigma series of computers — its third-generation line. In
jjfact, in that year (following IBM's announcement of System/360), the
17 i t
;[ Sigma family "occupied the attention of virtually every department in
:i
the company". (DX 981, p. 9.) It was announced in 1966, with the
15
13 ;
20 '
first of the series, the Sigma 7, announced in March of that year and
lithe second, the Sigma 2, in August. The remainder of the family was
21 -j .
j announced starting in 1967. (Palevsky, Tr. 3226.) By 1971, the Sigma
22
23
25
family included the 2, 3, 5, 6, 7, 8 and 9. (PX 5774, p. 13; DX
13400, p. 22; DX 13401, p. 19.) According to SDS, the Sigma family
delivered "more computations per dollar than any other commercially
i
1
I
ij -703-
•i
i
■j
1 ;
i
3 i
+i
5 !
i
5 :
i
7 I
t
1
3 J
9 !
10!
12 i
13 i
T4>> j' separate boxes at prices below IBM's, but a comparison could not be
15
17 !
available machine". SDS touted its versatility: "The impact of
Sigma, however, lies in its broad application for business, industry
and science as well as its ability to perform an almost unlimited
number of different applications at virtually the same time." (DX 981,
p. 6.)*
The Sigma Series was a response to the IBM 360 line. A
Sigma 7 press release, dated March 15, 1966, stated that Sigma "repre-
sents the first family of computers with an entirely new design since
the IBM 360 announcement" and that "Sigma 7 features a total capability
for both business and scientific data processing". (DX 52, p. 1.)
SDS attempted to set its prices 10 to 15 percent lower than IBM's
prices on the products that IBM had announced two years earlier.
(Palevsky, Tr. 3150.)** An effort was made to price each of the
made on the basis of the performance of the systems as a whole because
1C ! "then you get into the problem of what is a typical set of operations
jl and it becomes very complex to do". (Palevsky, Tr. 3269-71.) Appar-
it ently, SDS felt that a dollar price advantage was necessary to over-
'i come the obvious customer acceptance of System/360. (See Palevsky,
19 ;j
_ j Tr. 3149-50, 3176, 3270-72.)
20 ;!
21 || = ■
|j * For example, SDS contended that a Sigma "can simultaneously run
^2 ij an inventory control program together with a real-time process control
i| application. At the same time, 200 users at remote consoles through-
^ lout the country could be time sharing the central processor". (DX 981,
24 l
^ .« ** Currie testified that "generally" although "not always", SDS j
^ ;| tried to have somewhat lower prices than IBM for equivalent perform-
• ance on the order of 10 to 15%, and tried to have "an advantage over a \
i company like Univac" . (Tr. 15175-76.) !
;| I
'i i
I !
;j -704-
L I Palevsky testified that the Sigma computers compared to the .
Z <j 9300 "were more complex structurally. They were much faster, and some
3 ; of the computers in that line were compatible so that we had on a very
^\\ small scale something like the 360, that is , we had a number of
computers of various sizes that were program compatible," (Palevsky,
5 ; l Tr. 3165-66.) As with the 360, SDS "designed standard interface
7 j| units". It also "developed special programs which simplify the
design, engineering, and final assembly of various building blocks
or components into total systems." (DX 982, p. 12.)
For the Sigma series, SDS acquired Potter tapes, Control
Data disk drives, NCR printers, Teletype console typewriters and
Uptime card equipment* (Currie, Tr. 15507.) However, because periph-
eral equipment was viewed as a "critical element" in third-generation
computer systems, SDS had sharply increased its planning for internal
development and production of peripherals. In 1966, SDS began deliver-
ing its own magnetic tape units and had completed development of Rapid
;; Access Data files, which it called "two important peripheral products
17 i
for data storage that were completely designed and produced by the
i
j company". These products were expected "to enhance the capabilities
I of SDS computer systems". (DX 98 2, p. 10.) SDS stated its reasons
:i for undertaking peripheral equipment development programs internally
21 \
\ rather than acquiring independent manufacturers or continuing to
22 ''»
j purchase from suppliers as follows:
22 i
[ "First, SDS can realize a significant improvement in profit
24. 1 margins on equipment which the company produces internally.
!
25 }
j
:! -705-
3
9
ia
li
12
13
14-
13
IS
"Second, peripherals designed to complement the capabilities
of Sigma computers provide an additional competitive advantage
for SDS systems.
"Third, high reliability must be designed into the equipment
and quality control assured during production, thus minimizing
the cost for servicing faulty peripherals.
"Fourth, and most important, the technology is advancing too
rapidly to permit SDS to rely primarily on suppliers. The new
series of Rapid Access Data files is an example. Developed with
a considerable investment., SDS RADs are among the most advanced
secondary storage devices in the industry. The availability of
the various RAD models provides SDS with a significant advantage
in marketing Sigma computer systems." (DX 982, p. 10; see also
Palevsky, Tr. 3277-78.)
SDS provided "advanced software, including operating systems
for real time , batch , and time-sharing operations , FORTRAN IV and
COBOL compilers, assemblers" and various applications software,
including a library with "[m]ore than 1000 utility and mathematical
programs" for the Sigma family. (DX 49, pp. 2, 8.) SDS obtained
between 20% and 50% of this software — specifically assembly languages,
compilers , a Data Management System package , a linear programming
5 |j package and a communications package — from software houses such as
7 ij Digitek, Programmatics , Bonner & Moore, Informatics, Computer Usage
:l
3 .j Corporation, Computer Sciences Corporation, Dataware, CII , and
9 j Scientific Resources. (Currie, Tr. 15388-89.)
6 | At first, according to Palevsky, the Sigma series was
'1 | offered "to essentially the same market as before". Gradually SDS
4
2. '.I "started to market it for applications that were mixed scientific and
2 ; general data processing" (Palevsky, Tr. 3166) , and began to expand
!
It "\ into applications for general business and industry, including market-
.1
\z j ^ n< ? to business data processing customers. (Cohen, Tr. 14684-86.) By
i
i
i
I -706-
L
2
3
4.
this time, "changes in the technology" had begun to "blur" the dis-
tinctions between business and scientific data processing. (Palevsky,
Tr. 3137.) More and more customers started using a single computer
■ for both types of computation*; if one were to ask "say in '65 how
many installations used a single computer for both purposes and say by
'68 how had that grown ... I would guess that that had grown ten-
fold". (Palevsky, Tr. 3254-55.) Both because of that change and a
perception that the area SDS was focusing on was becoming too confin-
ing, SDS was "forced" to enter the "market" for business data process-
ing customers. (Palevsky, Tr. 3137.)
At the time of the Sigma 7 announcement, SDS issued a press
release stating:
3
7 i
3
9
ia
u i
12
12
L4-
15
IS;
17 j
IS (
; | (DX 932, p. 3.)*
IS
\ SDS was advertising its Sigma 7 computer system as "unfair i
20 I
1 to IBM . . . Sigma 7 does everything a 360/50 does. At a fraction of j
7f ! i
"Until now, explained SDS president Max Palevsky, computers
were generally built either for business data processing or for
solving problems of a scientific nature or for real time control
systems. 'But because of its advanced internal architecture,'
Mr. Palevsky stated, 'Sigma 7 is the only medium priced computer
that can deliver outstanding performance in any of these applica-
tions. ' " (DX 53, p. 1.)
Similarly, the Sigma 5 was "designed for operating real-time programs
■ simultaneously with general purpose scientific and business problems".
22 ■
| * In its 1966 and 1967 Annual Reports, SDS stated that although its
2- i Sigma series would perform business data processing applications, that
J did not mean that it was abandoning the customers on whom it had
25 heretofore built its business. (DX 982, p. 4; DX 983, p. 4.)
-707-
4 the cost. Sigma 7 is a little cheaper than the 360/50 and a good deal j
22. '"'» • !
i i
u
r !
faster. The combination gives Sigma a 25 to 65 percent edge in
cost/performance." (DX 54.) Cohen testified that the Sigma 7 "did
indeed" perform business data processing applications. (Cohen, Tr..
14631.) For example, the Harrison Radiator Division of General Motors
used Sigma equipment for its data processing requirements, which were
"principally" inventory control, which, according to Cohen, would be a
T jl business data processing application. (Tr. 14610-11.) Cohen also
I : \ testified that, looking only at the CPU, input/output processors and
5. ; main memory, the Sigma 7 could in fact do everything that the IBM
3 I 360/50 could do. (Cohen, Tr. 14622-24.)*
1 | According to an XDS sales guide (DX 50, p. 102/001-29), a
? 1 report based upon what was happening in the field, the 360/50, 360/40,
44, 65 and the 1800 were "competitors" of the Sigma 5 as late as 1972.
(Palevsky, Tr. 3232-33; see also Tr. 3185, 3228-29.) According to
Cohen, IBM's 360/44, 50, 65, 67 and 75, as well as the 1130 and 1800,
t|
^ !i were the IBM systems SDS "most commonly competed with". (Cohen, Tr.
■
7[
3
14555-56; PX 433.")**
ii
!l * However, when the array of peripherals and software available on
- I the IBM 360/50 was taken into account, Cohen was of the opinion that
.1 the Sigma 7 could not do all of what the 360/50 could. (Cohen, Tr.
•^ •] 14624.) Palevsky testified that the Sigma 7 was comparable to the
I 360/50 in terms of the hardware capability, but not in terms of the
- ! total system. (Tr. 3243-45.)
■j
!2 ii ** j n the early 1970s, the "prime competition" for the Sigma 5 was
•] the IBM 370 Models 135, 145 and 155; the DECsystem 10; the SEL (Systems
2 '■ Engineering Laboratories) 86/3 8 and the Univac 418-III. (Palevsky,
! Tr. 3231-32; DX 50, p. 102/001-29.) For the Sigma 6 and 9, the "prime
•- ] competitors" were the I3M 370 Models 135, 145, 155; the DECsystem 10
and the PDP 10, Models 1040, 1050, 1055 (dual processor); and the
?c
j Univac 1106. (Palevsky, Tr. 3247; DX 51, p. 103/001-13.) For the
-708-
z
4*
a
7
S
9
iai
Ll
12
13
14
15
15 i;
17 I
is ;
19
ZQ
Competition for the Sigma series included products from many
suppliers in addition to IBM. Cohen listed 21 companies (plus leasing
companies) as competitors to SDS/XDS "in the computer systems market"
during the period 1966 to 1972: IBM, Honeywell/ Univac, GE r CDC,
Burroughs, DEC, SEL, Modular Computer Corporation, Fischer & Porter,
Varian, Hewlett-Packard, Data General, Radiation, Inc., Harris,
Collins, Comten, Interdata, Electronic Associates ("on occasion") ,
EMR, RCA ("occasionally") and said there were "very likely" others.
(Cohen, Tr. 14600-09.)
Palevsky mentioned as competitors to SDS : IBM, CDC, DEC,
CCC (Computer Control, later acquired by Honeywell) , Univac ("occa-
sionally") , Burroughs ("rarely") and Honeywell (rarely, until the
latter' s acquisition of CCC) , and General Electric ("in certain
applications"). (Palevsky, Tr. 3166.) Engineering Associates, CCI,
NCR, ICL, EMR and COM were also competitors. (Palevsky, Tr. 3233; DX
50, p. 102/001-30.)
Cohen listed the "effective competitors" (he defined these
as companies which won 20% to 25% of the competitions in which they
;jwere engaged) (Cohen, Tr. 14723-24) of SDS in specific application
•.i
\ areas •
i In time sharing, he listed IBM, Honeywell (after the acqui-
j sition of General Electric 's computer business) , Univac, GE and
iSigma 8, the "prime competition" was the IBM 370 Models 145 and 155,
24 Ithe SEL 86/88, DECsystem 10, the Univac 1106/110 8, and the CDC 6200/
|6400. (Palevsky, Tr. 3238; DX 50, p. 102/001-30.)
25 j
-709-
)
!
l \
DEC.
In real time, he listed IBM, Honeywell, Univac, CDC,
DEC, SEL;
In seismic, IBM and> Univac?
In scientific batch, IBM, Univac, GE and later Honeywell
(after its acquisition of GE's computer business), CDC and DEC;
In communications, GE, Univac, IBM and Comten; and
In multi-use/multimode, IBM, Honeywell (after the GE acqui-
sition) , Univac, CDC and DEC. (Cohen, Tr. 14729-30.)
IBM competed with SDS as an "effective competitor" in every
application area, and the systems which competed with those of SDS
spanned the IBM product line: the 1130, 1620, 1800, 360/44, 360/50,
.•[360/65, 360/67, 370/145, 370/155 and 370/158. (Cohen, Tr. 14555-60;
.*!
7
3
PX 433.)
Indeed, Wright — IBM Director of Time-Sharing Marketing in
the period from 1964 to 1965 and Director of Marketing for Government,
|| Education and Medical Region from 1965 to 1969 — testified that SDS
; was among IBM's "principal competitors". (Tr. 12993.) IBM was well
; | aware of SDS during the mid- and late-1960s. On December 22, 1964,
"I Lear son reoorted to T. J. Watson, Jr., about the serious competition
i SDS posed to IBM's 360/40 and 360/50. (PX 1288, p. 2.) That concern
; intensified over the next several years. (See the discussion of the
??
£d
Model 44, pp. 412-13.) SDS also appeared as one of the nine "major
computer manufacturers" reported on in internal IBM reports on the
;-j financial results of certain of its competitors. (See, e.g., PX
IS j
~ 13451.)
;t
;| -710-
!
■i
L
Z
3.
4*
6
7
a
10
II !
e. The Merger. The Sigma family brought still more growth
for SDS. During 1966, which was described by SDS as "a crucial
transitional year", SDS offered two issues of convertible subordinated
debentures totaling $22,500,000 and retired short-term bank loans.
(DX 982, p. 3,) From, the fourth quarter of 1966 to the fourth quarter
of 1967, the year which SDS saw as a "critical" period "during which
the character of . . . future expansion" was "largely determined", SDS
doubled its production of EDP equipment. It stated that in 1967, it
"successfully completed its first product line transition, began a
major facilities expansion,, and initiated new product development and
cost control programs to sustain orderly growth". (DX 983, p. 3.)
££ j By 1969 , SDS had achieved that orderly growth and had reached
13
14-
15
IB
what would be, for SDS as an independent entity, the pinnacle of its
success . It told its stockholders that its international sales in
1968 had increased by more than threefold over 1967 and that "[t]he
sale of SDS oroducts outside the United States is expected to continue
,- !| to increase significantly through the company's increasing involvement
\
1S J in the various international markets". (DX 45, p. 12.) SDS stated
that it "ranks among the world's ten largest computer manufacturers"
| with assets of $113 million and more than 4,000 employees ( id. , p. 4),
19
I and described itself as "one of the world's largest suppliers of
ZL \ ■
•I commercial time-sharing systems". ( Id. , p. 6.)
22 :»
In 1969, SDS was acquired by the Xerox Corporation, a
company that had achieved "a position of eminence as a worldwide
;! enterprise" through the remarkable acceptance of the xerographic
:! -711-
L
r
r\
i
I \
3-1
3 i
!
3 !
.4. :
3 '«
7!!
si
•i
2. 1
•t
copier. (PX 5774, p. 6.) Xerox's revenues for the year 1968 (exclud-
ing Rank Xerox, Ltd., its British affiliate that marketed its products
abroad) were $896 million and its net income was $116 million (includ-
ing the income from Rank Xerox). (DX 13857, p. 3.) SDS was acquired
for approximately $980 million worth of Xerox stock.* (Palevsky,
Tr. 3195.)
[c j * The Xerox acquisition of SDS and the subsecuent activities of
) XDS are discussed below. (See pp. 1125-44.)
i
! -712-
1
2 1
*■;
a
7|
3
id
n
12.
13
I*
la ;i
17
48. Digital Equipment Corporation. Digital Equipment
Corporation ("DEC") was founded in 1957 by Kenneth Olsen and
Harlan Anderson (Hindle, Tr. 7318.),* both having previously been
**;' associated with M.I-T.' s Lincoln Labs where they worked on
Whirlwind and SAGE. (DX 13833, p. 5.) dec set up production
on one floor of a converted woolen mill in Maynard, Massachusetts,
with three employees. (DX 13858, p. 1.) Its initial capitalization
was $70,000, all of which was invested by the American
Research and Development Corporation — a Boston-based venture
capital firm. (Hindle, Tr. 7476.) Its first products were
laboratory logic modules — "printed circuit boards containing
components which are used to do logical functions in an
electronic sense: add, etc." — that were then used to test
and build other manufacturers 1 computers. (Hindle, Tr. 7318-19;
DX 13858, p. 1.)
The story of DEC is one of extraordinary growth
and enormous success in the computer business. From its
.- j beginning and throughout the 1960s, DEC achieved extremely
19 '.j j
„_ i * Winston R. Hindle, Jr. was the only witness from DEC. j
^° j Mr. Hindle joined DEC in 196 2. From 1967 through j
„ t : | his testimony in 1975, Mr. Hindle was Vice President and j
^* :{ Group Manager of the company with responsibility for numerous j
■j products within the DEC product line. His responsibility j
2— j encompassed the development, marketing, sales support, ;
j planning and financial areas. Mr. Hindle had served on j
— s j the executive committee of DEC, as well as the Finance and j
| Administration and Marketing Committees. (Hindle, Tr. j
24 } 7313-18, 7337.) As of 1979, Mr. Hindle was Vice President, |
^ j Corporate Operations. (DX 12323, p. 47.)
25 ■ • !
1 I
1 -713-
•;t !
t ! rapid growth by taking advantage of new technology and its
* I own research and development to manufacture an ever-expanding
3.! product line. DEC'S total assets grew from $5.7 million in
*|.: mid-1964 to $114. 8 million' in. mid-1970 . * (DX 511, p. 14;
5 DX 1384-S, p. 10.) DEC's profits after taxes went from $889,000
in 1964 to $14.4 million in 1970 (DX 511, p. 1; DX 13845, p. 10.)
Its worldwide EDP revenue grew from $4.3 million in 1961 to
$12-6 million in 1964, to $142.6 million in 1970, rising at
an annual compound growth rate of 44% per year. (DX 526.)
By virtue of the $70,000 investment in 1957, American Research
and Development had acquired 78% of DEC's common stock.
(DX 13833, pp. 5-6, 21.) In 1968, it sold 215,000 shares of
DEC stock for a gain of more than $26 million (DX 13834, p. 9)/
and in 1972, it distributed the remainder of its DEC stock,
valued at $382 million, to its shareholders. (DX 514, p. S;
DX 13835, p. 4.)
DEC acquired more and more space in its woolen
mill, and its original three employees were joined by many
others. It began expanding overseas in 1964. It formed
its first sales subsidiaries in the United Kingdom and
Australia in that year. In 1965, offices in Canada and
Germany were added. (DX 13845, p. 3; DX 13846, p. 3.) Sales
* Financial information for DEC first became publicly
available in 1964.
-714-
I
2
Z
offices in France, Japan and Sweden followed in 1966. (DX 13847,
p. 19.)* By 1966, DEC occupied 338,000 square feet in its original
location in Maynard, Massachusetts and employed 1100 people (DX 13847,
4» ! p. 3), had 24 sales offices in six countries, and had about 800
3"
e
7
a
9-
10
II
12
13
L4
15
IS ,.
i
is I
19 I
20 !
computer installations. (DX 517, p. 1.) By 1970 it had manufacturing
plants in England, Puerto Rico and Canada, as well as several in
Massachusetts, employed 5,800 people (DX 511, p. 3.), and had
computer installations in eleven countries. (DX 517, p. 2.).
The financing of this expansion required capital. In
1963, $300,000 was borrowed from American Research and Development
Corp. However, by the end of its 1964 fiscal year, DEC had accumu-
lated over $3 million in retained. earnings. (DX 13845, pp. 10-11.)
Retained earnings rose to $4.3 million in 1965, $15 million in
1968, $24 million in 1969 and $38.8 million in 1970. (DX 511, p. 15;
DX 13846, p. 12; DX 13979, p. 7.) DEC made its first public stock
offering in August 1966, raising $4,800,000. (PX 5026, p. 15;**
DX 13847, p. 3.) From 1968 to 1970 it had three additional public
offerings, raising a total of $63.5 million. (PX 4562, p. 17; DX
511, p. 17; DX 512, p. 11; DX 13979, p. 8.) A review of DEC'S Annual
£t j * In 1961, all revenue was domestic. In 1964, 91% of revenues
|j were generated domestically. By 1970, the domestic percentage had
22? dropped to 72%. (DX 526.)
23 I
24
25
** A more legible copy of PX 5026 has been marked as DX 13848.
-715-
I
z
a
Reports reveals that this outstanding record enabled DEC to meet
its financing needs with no reliance on long-term bank loans. As
Hindle testified, "Digital's expansion has not ever been limited by
the ability to raise cap ital . " (Tr. 7476.)
It is interesting to note in this connection that DEC
generally chose not to tie up its capital in financing leases for
7 | r its customers, feeling that it had "better ways to invest" its
SI money. When its customers wanted to lease, DEC would put the
9- j customer in touch with leasing agents who would "use their capital
10' I. and not Digital's capital". These "leasing agents" were organizations
11 .; "willing to finance a customer's computer over a period of time".
12.! When DEC did offer leases directly to its customers, it generally
i
13 sold them to financial institutions immediately thereafter.
■
IA, (Hindle, Tr. 7369-70.)
15 i DEC'S success was due in part to its commitment to product
I
15 |f development. Hindle testified that "Digital has through the years
i
i
U | spent between 8 and 11 percent of revenues annually on research and
IS I product development. ... We have felt that product development
i
ig • was a vital part of our success. The rapid advances of technology
20 | in the computer field have meant that we must keep abreast of these
2i | advances and incorporate them and understand them in our product
22 H lines as rapidly as we could." (Hindle, Tr. 7383-84.) That
23 |{ commitment paid off and, as Withington testified, DEC "always
24. ■ maintained a position of technological leadership or at least
25 ; currency with any significant competitor and always provided an
-716-
I
2
3
4*
5
&i
7
3
9*
10
II !
12!
1A !
15
16 J
17 !
IS j
19
adequate breadth of product line, maintenance and support".
(Withington, Tr. 56016.)
DEC's computer systems began with small, high performance
hardware offered with relatively little support to sophisticated
customers capable of providing themselves the software and services
for the application of those products to their needs. Palevsky of
SDS described the early DEC computers as being similar to the small
and medium-scale 900 series computers marketed by SDS principally
for real-time and scientific applications, also in the early 1960s.
(Palevsky, Tr. 3133-36.) Over time, DEC's products grew in capacity
and capability and DEC expanded its customer support, its marketing
and its software and service offerings . That process has continued
and DEC now offers one of the broadest product lines in the industry
and markets it to the whole spectrum of EDP customers. (See below,
pp. 717-31, 989-92.)
DEC's first computer, the PDP 1, was delivered in 1960,
and it was followed by the PDP 4 and 5 in 1962 and 1963, respectively."
From 1964 through 1970, DEC also introduced the PDP 6 (1964), PDP 7
!l (1964), PDP 8 (1964), PDP 8S (1966), LINC 8 (1966), PDP 9 (1966),
„„ ! PDP 10 (1967), PDP 81 (1968), PDP 8L (1968), PDP 12 (1969), PDP 14
20 j,
I (1969), PDP 15 (1969) and PDP 11 (1970). All of these DEC computers
»• it !
i were classified by Hindle as "general purpose computers", except
23 !
24 !
25
the PDP 14.* (Hindle, Tr. 7321-24, 7327, 7388; PX 377-A. ) As we
i
* The PDP 14 was not a general purpose computer "because it has I
a program which is preset prior to its delivery, and then it only ;
operates on that same program when used by the customer". (Hindle, I
Tr. 7327.) j
i
i
-717-
I
2
Z
4-
5
6
7
3
9-
10
11
12
13
15 i
16
17
IS
19
shall see, each of DEC's successive product announcements expanded,
the breadth and capabilities of its product line, contributing to
DEC's phenomenal success.
a * PDF 1/ 4, 5 and 7. DEC's first computer was the PDP
1, delivered in 1£6Q (and withdrawn in 1963 or 1964) . (Hindle, Tr.
7318-19, 7321; DX 507, p. 10.) The PDP 1 was "an outgrowth of the
technology that was incorporated into the line of logic modules ,
although there, was a completely separate design from the logic
modules in our product line prior to that". (Hindle, Tr. 7319.)
The original purchase price of a PDP 1 was in the neighbor-
hood of $125,000 to $243,000. (DX 13858, p. 2; Plaintiff's Admissions,
Set II, 1111 240.2, 371.3(d).) It was designated "PDP" (Programmed
Data Processor) , a nomenclature that DEC used throughout the 1960s
for its products, because "EDP people could not believe that in
1960 computers that could do the job could be built for less than
$1 million". (DX 13858, p. 2.)
At Stanford University, Professor John McCarthy used a
PDP 1 to conduct some of the earliest research on time-sharing in
the early 1960s. (Feigenbaum, Tr. 29531-32, 29535-36; DX 13858, p.
20 :' 3.)* A PDP 1 was also used, in conjunction with an IBM AN/FSQ-32
2i |i
22 i; * The PDP 1 "operated primarily independently", but was "linkfed]
| through a disk" to an IBM 7090. " [O] ccasionally there would be a
23 ;! lash-up which would exercise LISP on the 7090, which is a system
developed ... [at Stanford] that would communicate through the
^ disk to the PDP-1." (Feigenbaum, Tr. 29532-33. ^
-718-
1 ; | computer, as "the major input/output vehicle for the various remote
2 J devices" in a "general -purpose time-sharing system" at the System
3-i Development Corporation (SDC) in June 1963.* (DX 7622, p. 3.)
4 I That time-sharing system was produced "under the sponsorship of
5 | ARPA[**] and. . - • utilized ideas developed at both Massachusetts
$ !i Institute of Technology and Bolt, Beranek, and Newman, as well as
7 i| some original techniques". The various remote devices used as part
3 ;! of this system included "Teletypes . . . and other computers" that
g I) could be run "from within SDC, and from the outside". PDP l's, as
i
1Q i well as the CDC 160A and the IBM 1410, were also expected to be used
it
H J at remote stations as part of this system. ( Id. ) In 1962, the
■i
12 |l Atomic Energy Commission's Lawrence Livermore Laboratory selected a
ii
13 || PDP 1 over a 1401 proposed by IBM to perform what it described as
ii
14 i "scientific and engineering" applications. (DX 2992, pp. 18,
ii
15 ;! 1113.)
15 : | DEC introduced the PDP 4 computer system in 1962, and the
17 l| PDP 5 in 1963. (Hindle, Tr. 7321.) The PDP 5 was offered with
,a 1 keyboard-printer, paper tape reader and punch and a software package
19 !
i
2q 1 * "Time-sharing, in this case, means the simultaneous access to
i a computer by a large number of independent (and/or related) users
2i ! and programs." (DX 7622, p. 3, emphasis in original.)
22
** ARPA is the government's Advanced Research Projects Agency,
which was established in 1958 and whose, "primary mission ... is
2- ■; to support research and development of advanced projects which have
potential value to the Department of Defense". (Plaintiff's
Admissions, Set I, 1MI 1.0, 2.0.)
±3
24
25
-719-
i
f
z
T
3
q :
including FORTRAN. (DX 13928 .) "Although the term •minicomputer'
wasn't used during the time of the PDP- 5," DEC would have considered
the PDP 5 a "minicomputer". (Hindle, Tr. 7325.)* According to the
1964 DEC Annual Report, the- PDP 5 was used in "numerous applications
in physics , biomedicine, industrial process control, and systems
applications " . (DX 13 845 , p . 8 . )
In the early sixties, DEC'S research and development
effort produced the "flip chip" modules (DX 13845, p. 3) which,
like IBM's SLT, combined printed circuits with discrete components
( id . , p. 2) and which made possible the introduction in 1964 of the
smaller, faster and cheaper PDP 7 and 8 to replace the PDP 4 and
5.** An IBM Win and Loss Report for August 1964 reported competition
between the DEC PDP 4 and the IBM 1401 and between the PDP 7 and
^ : . the 360/30. (PX 3630, p. 6.) The Atomic Energy Commission selected
a PDP 7, installed in January 1967, over a 360/30, as well as systems
i
bid by CDC, SDS, Univac and Honeywell. (DX 2992, p. 49.)
.a ■
j ;|
[\ * Hindle testified that the term "minicomputer" (which "[pleople
•S :j in the industry started to use ... in the middle 1960 's") "is not
;j a precise term", but rather has "a broad range of definition [s] " as
•S :} used in the industry. Hindle* s "own view" of a minicomputer is a
j system "priced at less than $50,000". (Tr. 7325.) By this he
£G j meant that "the smallest available configuration could be con-
:j figured for less than $50,000". (Tr. 7453.)
U ■!
j ** Withington commented in 1964 that the rapid reduction in
22 ; | manufacturing costs of high-speed circuits enabled DEC, among
j others, "to enter the computer market with low-priced computers
ZS : j of high performance". (PX 4829, p. 31.)
24. 1
-720-
I
2
3
4
5
6.1
7
3
9-
10
11
12 l
13
14
15
15 ;
17|
i
is I
19
20
b * PDP 6. The PDP 6 was first delivered by DEC in 1964.
(Hindle, Tr. 7321.)*" DEC'S Annual Report for 1964 described the
PDF 6 as "an expandable system which can start as a very basic
configuration and grow through the addition of processor, memory
and input-output options into a major computation facility equivalent
to the largest commercial systems currently offered". (DX 13845,
p. 6.) At the time, the PDP 6 was the largest of DEC'S computer
systems, ranging in price from $350,000 to $750,000. DEC described
it as "equivalent to the very large computers used by scientific
laboratories". The PDP 6 was used in "large data processing assign-
ments", including Brookhaven National Laboratory, the Rutgers
University Physics Department, the Universities of Aachen and Bonn,
Germany, the University of Western Australia, Lawrence Radiation
Laboratory, United Aircraft Corporation, Applied Logic Corporation,
Yale, MIT's Laboratory for Nuclear Science, the University of
Rochester, Stanford University and the University of California at
Berkeley. (DX 13845 , p. 3; DX 13846 , p. 8; DX 13847, p. 7.)
Also, the PDP 6 was "designed for time-shared use" and
'(DEC bid a multiprocessor version of it to MIT's Project MAC, one of
the earliest and most important experiments in the use of time
ij sharing. DEC was "in among the finishers" for this award, who
| included: CDC, bidding a 6600; IBM, bidding a 360/50; GE, the
22
24;
25 i
* The PDP 6 was withdrawn in 196 7, when it was succeeded by the
PDP 10 which "incorporates all the features of the earlier machine".
(Hindle, Tr. 7321; PX 5026 (DX 13848, p. 3).)
-721-
Li
i
z\
l!
t
i
!
i
7i
8 |
i
Si
-I
u|
i
12 j
13
14
15
16
17
13 ;
19 I
20 !
winner of the contract with a modified 635; and RCA, with a 3301.
A $1 million PDP 6 was chosen by Project MAC as a peripheral processor
for the time-sharing system. (PX 2961, pp. 1, 3; DX 13845, p. 3.)
An IBM Win and Loss Report for August 1964 reported additional
«
competition between the PDF 6 and the IBM 360/50. (PX 3630, p.. 6.)*
At the Lawrence Berkeley Laboratory of the Atomic Energy
Commission, a PDP 6 was bid against an IBM 360/70 and a CDC 6600,
among others. A CDC 6600 was installed in January 1966. (DX 2992,
p.- ID
At Lawrence Livermore Laboratory , two PDP 6 ' s were acquired
in 1965-*66 and used as control computers in the OCTOPUS network.
Because Lawrence Livermore required a faster memory than DEC could
supply, it solicited bids for add-on memory, receiving bids from
five companies in addition to DEC, and awarding contracts to Lockheed
and Ampex. (DX 4572.) That acquisition was an indication of
things to come: there was later, primarily in the 1970s, substantial
development of DEC plug-compatible equipment, in part reflecting
the great popularity of DEC computer systems. (See, e.g., Hindle,
Tr. 7422-23, 7444-45.)
c. PDP 8. The PDP 6 procurements were prestigious, but
•I
^ :j DEC's financial growth was more affected by the PDP 8 series of
21 ;| •
li
22 :;
| * The United Aircraft Research Laboratories offered computation
22 i services with equipment consisting of an IBM 360/50 system, a
;! Univac 1108 system and a "DEC PDP -6 Time-sharing system" including
24 it "a paper tape reader, and both drum and magnetic tape auxiliary
;■ storage" and accessible "through most standard terminal units".
25 ; (DX 7506, p. 44.)
i
! -722-
I
2
Z
4-
5
a
7
3 !
»i
u |
12 (
i
u !
I
i
IS j
17 j
13 ;
is- ;i
computer systems — one of DEC ' s most successful. The PDP 8 was
first introduced in 1964 as a replacement for the PDP 5. Due to
the use of integrated circuits , the PDP 8 was four times faster
than the PDP 5 at two-thirds the price. (Handle, Tr. 7321; DX
13845/ p. 3; DX 13846/ p* 3.) Using the price/performance improve-
ments made possible by the "FLIP-CHIP circuit modules and automated
l| production techniques", the PDP 8 opened new market opportunities
for DEC (such as typesetting) and expanded its base of scientific-
oriented users. (DX 13846 , p. 3; DX 13847 , p. 8.) The PDP 8 was
offered with "disk storage units, terminals, tape units, line
printers, cathode-ray tube display units". (Hindle, Tr. 7334.)
Perlis testified that the PDP 8 "generalized very nicely
to other machines and it itself gave birth to a whole line of
offspring" (Perlis, Tr. 1877); the "parent" itself "received a
remarkably immediate acceptance". (DX 13847, p. 8.) As a result
of the greater than expected demand, DEC expanded its manufacturing
facilities in 1965. By mid-1966, over 400 PDP 8's had been installed*
(DX 13846 , p. 3; DX 13847 , p. 8.)
The various other members of the PDP 8 family in existence
at the time of Hindle 's testimony in November 1975 were introduced
20 :
j] from 1966 (the PDP 8S) through 1974 or 1975 (the PDP 8A) with the
21 '«
j; later members of the family still in delivery. (Hindle, Tr. 7322.)
22
24
25
After the original introduction of the PDP 8, each new member of
j the PDP 8 family was introduced at a lower price because of "changes
in manufacturing technology, semiconductor prices, and peripheral
-723-
prices as purchased from our vendors". (Hindle, Tr. 7347-48.)*
The systems of the PDP 8 family were software-compatible (Hindle,
Tr. 7421) with DEC providing three "general purpose operating
systems" for them. (Hindle { Tr, 7346-47.) It was "generally
true"'/ although "not . . • . one hundred percent true", that the same
peripherals could be used on all systems in the family. (Hindle,
Tr. 7421-22.)
First deliveries of the PDP 8 went to such organizations
as Stanford Research Institute, Harvard Medical School, Massachusetts
Institute of Technology and the University of Wisconsin. The PDP 8
was also offered to the newspaper and book publishing fields for
automatic typesetting (DX 13846, p. 6) and to laboratory, industrial
and educational users. (Hindle, Tr. 7331.)
In addition, approximately 30 to 50% of the PDP 8 family
was sold OEM,** some as processors alone, some as processors plus
memory, and some as total systems. (Hindle, Tr. 7330.) When sold
OEM as a processor only, the purchaser would acquire "from another
manufacturer the appropriate devices necessary to perform input or
|! output functions", with either the customer or DEC providing "inter-
im
2
3
4j
5
a
7\
3
9
10
11
12
13
14
15
17 I
13 !
19 i
h
jj * Hindle testified that in pricing its products, DEC takes
2 * ! i several factors into account: (i) the computer systems DEC believes
;! will be competitive with the one being priced, (ii) conditions in
22 ;j the segments of the market into which the product is expected to be
i| sold, (iii) manufacturing and support costs, (iv) expected profit.
23 : j (Tr. 7337-38.)
24 ; ** OEMs are systems vendors or manufacturers which incorporated
l| the PDP 8 processors into their systems or products. (Hindle, Tr.
" ; | 7330-31.) DEC generally charged OEM buyers and end-user buyers the
j same prices. (Hindle, Tr. 7 34 8.)
" -724-
.!
■ i
;i
I
2
5
el
i
7\
8 !
9-!
t
10
I
n !
facing services" for the input and output devices. (Hindle, Tr.
7330-31.) Since 1964, these OEM purchasers have marketed PDP 8
systems for business data processing. (Hindle, Tr. 7332.)
Business data processing applications that are performed
on PDP 8 computer systems include "invoicing, accounts payable,
inventory control, order processing" and others. (Hindle, Tr.
7389.) In addition, PDP 8' s have been used for "real time data
collection from instruments", "to assist in the teaching process",
in "industrial control applications", including "the automation of
industrial process controls, the collection, analysis and reporting
of quality control data, test data, material handling data", in
J2, j "commercial typesetting applications" such as "copy editing, in
hyphenization and justification" and "setting classified advertise-
ments", in "data communications applications" such as "message
switching, data multiplexing, data concentration" or "front end
processing". Different users use the PDP 8 to perform different
applications and in some cases the same user might use the same PDP
8 computer system to do, for example, both business data processing
and industrial control applications. (Hindle, Tr. 7389-91.)
DEC itself, prior to 1972, had largely been "unsuccessful"
23
24
25
tions because it had not worked "on the packaging to make the
product suitable and attractive to the business data processing
customers". (Hindle, Tr. 7489-90.) Then, in 1972, DEC introduced
the Datasystem 300, an adaptation of the PDP 8 specifically designed
-725-
L j for business data processing. The "primary difference" between the
i
Z-' { PDP 8 and the DEC Datasystem 300 was that the latter was packaged
3 [ in a different type of console and used a business-oriented language
4 1 called DIBOlr. The Datasystem 300 was offered with the same peripherals
i • '
f j as the PDF 8.. (Hindle, Tr. 7333-34.)
i
I
$ I During the period prior to 1975, the PDP 8 competed with
i
7 I IBM's 360 and 370 computer systems when used as part of larger DEC
3_ ; computer systems like the PDP 10 and singly against the IBM System/3,
g. | System/32 and 1130. (Hindle, Tr. 7341, 7442; PX 377-A.)* IBM r s
John Akers recalled losing to a PDP in 1966, when he bid a 360/20
for a typesetting application at Worcester Telegram. (Akers, Tr.
j^ | 96713.) The PDP 8 also had the ability to perform terminal or
12 i input/output applications as part of a computer system with an IBM
!
^j GPU or some other manufacturer's CPU. (Hindle, Tr. 7394.) DEC
i
je ! considered the products of IBM when setting the price for the
i
.- <\ Datasystem 300. (Hindle, Tr. 7338-39, see Tr. 7341.)
1T ; By any standards, the PDP 8 was a successful and significant
product. At the time of Hindle 's testimony in late 1975, between
10
11
1S „
20 ;
30,000 and 40,000 PDP 8's had been sold (Hindle, Tr. 7329), and the
* Hindle testified that in drawing up PX 377-A (which was offered
in evidence "for an illustrative purpose" and "to assist Mr. Hindle
(j in testifying about each of the products listed thereon" (Tr. 7320-
2- ij 21)) and in testifying to competition generally, he meant "one for
j| one competition", i.e. , a situation in which both DEC and DEC's
^ : competitor will bid for a product with the same price and the same
|j performance to do the same job or where DEC "would bid two or three
2-" ;j of our products to compete with one of the products from the other
company". (Tr. 7414-17.)
21
25
-726-
2
Z j
a
7
3
9
10
U
12
"13
PDP 8 was being marketed "to a wider variety of customers", including
"communications customers" and "business data processing customers".
(Kindle, Tr. 7331-32.)
<*• B®P 10'- In 1967, DEC introduced the PDP 10.* The
PDP 10 was the first of a family of computer systems later marketed
as the "DEC System 10". (Kindle, Tr. 7324.) Kindle testified that
the PDP 10 could do "all of the different kinds of applications
that, are performed by the PDP 8" as well as additional applications
characterized by DEC as "non-business computation applications",
such as the manipulation and analysis of "scientific, engineering,
or numerical data".. (Tr» 7391-92.) The PDP 10 was "introduced to
serve . . . laboratory users, industrial users, education users, a
class of users we call the data service industry. . . . This is a
• . f class of computer users who purchase— or lease or rent — computer
.- jl equipment and then offer various kinds of services to clients."
^m* L
i (Kindle, Tr. 7359.) When introduced, it could have been purchased
"for prices in the vicinity of $450,000, all the way up to configura-
tions which would be a million and a half dollars". (Hindle, Tr.
7359.) It was not actively marketed on an OEM basis, but generally
i
was sold directly to end users. (Hindle, Tr. 7358.) !
: !
i
21 \ ~ |
j * Using his definition of "minicomputer" (a computer system in j
22 : | which "the smallest available configuration could be configured for j
; j less than $50,000" (Tr. 7453)), Hindle classified all DEC computer j
23 j systems introduced during the 1960s as "minicomputers" except for ;
j the PDP 10 , which was too large and too expensive to meet that
24. '"} definition. (Tr. 7325-27, 7358-59.) !
i i
25 ; ■ ■ j
! -727-
17 j
IS ;
19 ;
2Q :
2 I
i
I
I
*l
i
5 I
»
i
6 j
7 |
a!
* I
10-1
11
12
13
L4
15
16
17
IS
In 1969, DEC described the PDP 10 as follows:
"serving business, industry, and science in a multitude of
installations throughout the world- They keep track of bubble
chamber events in physics laboratories, analyze blood chromosomes,
work in banks, teach in high schools and universities, and per-
form a myriad of other, tasks. New applications are constantly
appearing and current applications steadily grow. Customers find
new approaches, add new equipment, develop more software.
Systems designed solely for real-time tasks often expand to
include program development or business data processing. The
applications described here demonstrate the PDP-10's inherent
flexibility." (DX 519-B, p. 7.)
The PDP 10 competed on a "one for one basis" with the IBM
360 r 370 and System/3 computer systems. (Hindle, Tr. 7442; PX 377-
A.)* In establishing the price of the PDP 10, DEC looked at the IBM
360 series, specifically the Models 30, 40 and 50, as well as systems
offered by Honeywell, General Electric and Scientific Data Systems.
(Hindle, Tr. 7361-62.)
The DECsystem 10 was announced in 1971 as a family of
systems "spanning virtually the entire large-scale range" and was
based upon the PDP 10 processors.** (DX 512, p. 1; DX 522, p. 3.)
(The DECsystem 10 is discussed in the section of this narrative deal- ■
.- :i * This competition is demonstrated by the procurements of govern-
i| ment agencies. For example, at the Atomic Energy Commission, a PDP 10
2 i| was proposed in competition with an IBM 360/44 and a SEL 810A. The |
;■! SEL 810A was selected and installed in December 1968. Earlier that |
-^ ij year, the Commission selected a Sigma 7 bid against an IBM 360/50 and j
: a PDP 10. On another occasion, a PDP 10 was successfully bid against j
| a 360/50 at the Commission and installed in late 1969. The Depart- j
^ jj ment of Health, Education and Welfare chose a PDP 10 over a proposed j
360/40 in 1969. (DX 2992, pp. 73, 86, 118, 858.) I
23 •
24 3
25
** The DECsystem 10 peripherals embodied some improvements compared
to the peripherals offered with the PDP 10. In particular, there
!| was a better quality printer and a disk drive with a removable disk
'i pack. (Hindle, Tr. 7362-63.)
;| -728-
u
l\
i
it
L
Z
z
4-
5
a
ing with the 1970s.) Both the PDP 10 and the DECsystem 10 were
highly successful. By 1974, over 400 such systems had been installed.
(Kindle, Tr* 7380; DX 525, p.. 10.)
e * ?PP 15 » The PDP 15, introduced in 1969 (Hindle, Tr.
7323) , was originally "marketed to laboratory users , industrial
users, education users", but with the addition of another programming
7; language ("MUMPS")— a "business-oriented language" — it was also
3
10
11
12
13
marketed for business data processing applications starting in the
early seventies. (Hindle, Tr. 7377.) "Representative applications"
performed by the PDP 15 were "nonbusiness computation applications,
business data processing applications, real time data collection and
instructional computing applications as well as industrial control
applications". (Hindle, Tr. 7441; PX 377-A.) Hindle testified that
the PDP 15 competed "on a one for one basis" with the IBM 1130, 1800,
£5 j System/7, 360 and 370 computer systems. (Tr. 7441; PX 377-A.) By
♦ fi ' late 1975, between 800 and 1,200 PDP 15s had been installed. (Hindle,
y- j Tr. 7380.)
e ., f. PDP 11. As with the PDP 8, the PDP 11, introduced in
a j! 1970, was the designation for a family of computer systems. (Hindle,
| Tr. 7323.) At the time of introduction DEC expected to market it "to
j the entire group of users . . . described for the PDP 8, which would
\ include laboratory users , education users , industrial users , engineer-
„_ J ing users, [and] communications users". As with the PDP 8, between 30
« i
and 50 percent of the PDP lis were sold to OEM purchasers who wrote
24 )
25
applications programs and offered the PDP 11 for business data process-
-729-
ing applications.* . (Hindle, Tr. 7349-51.) But, DEC preferred to
market the PDP 11 as. a system rather than just the processor. (Hindle,
Tr. 7349.) In 1972:, the DEC Datasystem 500 series was introduced for
marketing to business data processing customers. DEC took the PDP 11,
added software capabilities, including BASIC, and put it in a "sepa-
rate type of package" that looked different to the user. "But other
than that, there were no significant differences ►" (Hindle, Tr. 7351-
52, 7355-57.)
Hindle testified that the PDP 11/15 and 11/20, smaller
members of the PDP 11 family r competed on a one-for-one basis with the
IBM 1130, System 3 and System 7 computers, and also with the 360 and
370 in configurations which included the PDP 10. (Tr. 7441 t PX 377-
A. ) Larger members of the PDP 11 family, like the 11/45 and 11/70,
competed on a one-for-one basis with IBM 360 and 370, as well as
System 3, System 7 and 1130.** (Tr. 7414-15; PX 377-A.) Like the PDP 8,
the smaller PDP 11 computer systems might "be used to perform terminal
or input/output applications" as part of computer systems whose main
I
V
7
5
7 \
3
3
a
1!
2
3
.4
J 1
.3]
;| "An OEM customer of ours . . . would then hire and train
^ ,j and use programmers, who would write applications programs for
;i the PDP 11 system for that particular problem application that
;j he had identified as a marketing segment, and then would
jj proceed to sell the combined PDP 11 system with the application
programs that he had designed to the end user." (Tr. 7351.)
* Hindle described the OEM marketing as follows :
?*r
' »
23 H ** The PDP 11/45 and PDP 11/70 were announced in 1971 and 1975,
respectively. DEC does not consider the PDP 11/70 to be a "mini-
2A 1 computer". (Hindle, Tr. 7323, 7325-27.) Even some PDP ll/45s have
i been configured into systems with prices as high as $250,000.
25 j (Hindle, Tr. 7456.)
-730-
L
Z
3
4
5
7
3
10
II
IZ
13
7 ^ p 1970 DEC had "introduced many new peripherals including those of our
CPU was manufactured by IBM or by manufacturers other than IBM or
DEC. (Hindle, Tr. 7394.)
In. pricing the products in the DEC DataSystem 500 series,
DEC looked at the prices of the IBM System 3, System 7 , System 32, 360
and 370 systems.. (Hindle, Tr. 7354.)
The PDP 11 family has been a highly successful and signifi-
cant product line, but its story, like that of the DECsystem 10,
unfolds in the 1970s.
g. Peripherals and Software. The rapid proliferation of
DEC'S product line during the 1960s extended to peripheral and soft-
ware ' of fe rings as well. Although as late as 1969 Memorex was market-
ing disk file products to DEC on an OEM basis (Spitters, Tr. 42067-68)
with 1969 sales of approximately $5,000,000 (Spitters, Tr. 42072), by
15
15 _
17 |
is ;
19
20 ;|
21 |
22
23 !
24
25
own internal design and manufacture, such as disks, paper tape,
| t DECtape, display systems, and real time interface equipment". (DX
517, p. 2.) DEC's 1970 Annual Report proclaimed that "[i]n order to •
expand the capabilities of its computers, DEC provides a wide range of
I peripheral equipment", including large magnetic tape systems, storage
drums, teletypes, high speed paper tape readers, card readers and
punches, line printers, incremental plotters, digital-to-analog
converters and various controllers. (DX 511, p. 10.)
DEC had also worked on software, introducing new software
features " [w] ith each mainframe that is a new version of a previous
machine". (DX 517, p. 2.) It had provided "DIBOL" , a "business
-731-
1
2
3
4*
5
6
7
3
10
11
oriented language " for the PDP 8, and added COBOL to the PDP 10. (DX
517, pp. 3, 5..)
in 1967 it developed application packages called " Computer -
packs" which from?: 1967 to 1$69 were marketed together with the DEC
hardware at no separate charge. (Hindle, Tr. 7426.) The "Computer-
pack" was actually a complete turnkey system for users desiring
systems that required a minimum of programming and computer experience
DEC merely added an application package to a PDP 8 and marketed
the result as a "Computerpack" . For example, the "Quiokpoint-8" was
offered for numerical control tape preparation; the "Communic-8" was
offered for data communication applications; the "Time-Shared- 8" was
t5 j offered for general purpose time-sharing applications; and the "LAB— 8"
13
15
was offered for nuclear magnetic resonance spectroscopy applications.
l±\ (DX 6868, pp. 7-10; DX 10776, pp. 8-13.)
h. Competition. DEC ' s approach to the market was different
^g j| than that of IBM. DEC in the 1960s offered fast, inexpensive hardware
17
13
with less versatile and generalized software and service than that
offered by IBM. As shown in PX 3 77 -A, DEC marketed most of its
• a :! machines announced in the 1960s to "experienced" and "moderately
'I
2q ! j experienced" users . Perlis described the PDP systems as they were
2«r ij perceived in the university environment: "[I]t was generally felt
-«. i! that . . . PDP systems . . . for delivering the same amount of work,
i
-- |j were cheaper than the I3M systems." He estimated that the PDP 10 was
_ A I about 20% cheaper than a 360/50 because of the "attendant staff of
« ' operators, people to handle the variety of software that is used . . .
-732-
i
■I
and so forth" associated with the IBM 360 Model 50 while PDP 10' s were
operated "without any staff whatsoever in attendance on the machine
during its period of operation, which runs 24 hours a day, seven days
a week".* (Tr„ 1976-77.) In this respect, the bundle of services
associated with the IBM 360 line provided an opportunity for DEC to
obtain a price advantage with users who did not want or need those
services. Thus, as is illustrated by the discussion below, DEC
offered hardware and software more tailored than the generalized
System/360 to enable users to perform one or a few applications in a
decentralized way rather than on a central IBM computer.
Competition from DEC was felt within IBM in the 1960s.
Wright, who was a Director of Marketing in IBM's Data Processing
Division in the 1960s, included DEC on his list of "principal competi-
tors" during the 1964 to 1969 time period. (Wright, Tr. 12993.)
Similarly, when Rooney was employed by IBM in the mid-1960s as a
Branch Manager in New York, DEC was competing in the "marketplace" for
the "manufacture and marketing of systems for commercial or scientific
usage". (Rooney, Tr. 11733.) Akers recalled meeting DEC in three
different situations in which he was personally involved in the 1960s
Tr. 96713-14.) He studied DEC both as a salesman in Vermont and a
22
23
24
25
* Of course, with IBM "the user receives an enormous amount of
service, an enormous amount of software, very good maintenance and for
many users that is well worth the 20 per cent difference". (Perlis,
Tr. 1978.)
-733-
1
2
Z
5
S
7
3
9
10
11
12
13
14
15
IS
17
IS
marketing manager in Boston (Tr. 96679) and, when in the New York
Media branch office, found that IBM had "a good deal of competition
from the Digital Equipment Corporation". (Tr. 96680.)
The competition between DEC and IBM was not only on a "one~-
for-one" basis. As Hindle explained, " [i]t would be possible that in
a given computer application a customer could choose one powerful
machine to do the job or could choose several less powerful machines
and decentralize the job. In that type of situation we would have one
machine competing with several machines from a different manufacturer."
Such competition would arise, for example, because " [considering the
total system cost of both software and hardware, a distributed network
of smaller computers can often be a cost-effective alternative to the
single, centralized computer". Alternatively, as Hindle said, ,r [i]t
is possible to have several smaller computer systems which are not
interconnected electronically" competing with a single larger computer
system. (Tr. 7415-17.)
Similarly, the IBM Commercial Analysis Department described
this competition in the Quarterly Product Line Assessment for the
1$ ;| first quarter of 1970: "Mini-computers affect IBM's business poten-
20 : j tial by implementing one application out of several possible applica-
2x I! tions in a prospect's business." And, according to the report, the
22 *•; application selected for the minicomputer was frequently the applica-
23 ! tion having the greatest economic justification. The off-loading of
24 ; ! that application could eliminate the opportunity for IBM to supply the
-25 customer with a "larger and more comprehensive computer installation".
-734-
L
Z
z
4-
5
**■
s >
7J
i
al
9-
10
II
EL
13
14-
15
IS
17
IS
»i
20 :!
* 'I
\
22)
1
23 J
4^ |
(PX 2567, p. 186.) The report also commented on the success of
minicomputers :
"Hani-computers have established a substantial base and
continue to widen their base each year. Digital Equipment
Corporation is now ranked #3 in total CPU's representing
about 5,600. units (8%)- out of the 70,000 total domestic
CPU's reported for year end 1969- 'by Diebold." ( Id. )
Other companies also met DEC as competition in the
1960s. Palevsky testified that DEC competed with SI>S ' s Sigma
Series. (Tr. 322S-29. ) Honeywell management believed that Honeywell
systems competed with systems from DEC as well as those from General
Electric and Hewlett-Packard.* (Binger, Tr. 4593-94.)
DEC entered the seventies a large and profitable company
with a successful and popular product line. Its 1970 fiscal year
revenues were $135.4 million with income before taxes of $25.5
million. (DX 511, p. 1.) It had some 500 computers installed in
the Federal government, or almost 10% of the total number, making
it the third ranked supplier in terms of numbers of computers
(behind IBM and Univac) . (DX 924, p. 6.) But, despite the
impressiveness of those indicators of success, they were but
small fractions of the DEC that was to emerge in the next decade.
* Spangle included DEC as well as Sperry Rand, NCR, Burroughs
25 ] and CDC in a list of Honeywell's competitors. (Spangle, Tr.
4933-34.)
; i
;! -735-
3
I *
T
49.. AT&T .. Despite the continuing restrictions of its 1956
Consent Decree, AT&T expanded its offerings of computer-related
products and services during the 1964-1969 period. Its U.S. EDP
: revenues, a*s a result, grew from- $125.6 million in 1964 to $477.75
fj| million by 1969- ' (DX 8224, p. 133.)
As it did in the 1950s, AT&T competed in the computer
industry in at least two ways during the 1960s. The first involved
Western Electric ' s manufacture and marketing to the Bell System
operating companies of stored program controlled electronic switching
systems and automatic intercept systems. Because of the Bell System's
enormous size and the fact that the Bell operating companies are free
to and do in fact buy EDP products and services from non-Bell
affiliated companies, this is a very important source of business to
computer vendors who vie with Western Electric for the business of
the Bell System. (See DX 5945, Dunnaville, pp. 6-8 and discussion
below.) During the 1960s equipment developed and manufactured, by
AT&T competed for- the business of the Bell System with the equipment
of other EDP vendors, including IBM. (Id.)*
*3
17
19 ii
! The second form of competition is AT&am